diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,22786 @@
+{
+  "best_metric": 0.8992551940415523,
+  "best_model_checkpoint": "videomae-large-cctv-brawl_extended_v2/checkpoint-25140",
+  "epoch": 14.066295264623955,
+  "eval_steps": 500,
+  "global_step": 37695,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.326259946949602e-09,
+      "loss": 0.6671,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.652519893899204e-09,
+      "loss": 0.7063,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.978779840848806e-09,
+      "loss": 0.6964,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.305039787798408e-09,
+      "loss": 0.684,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.631299734748011e-09,
+      "loss": 0.7722,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.957559681697613e-09,
+      "loss": 0.6667,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.283819628647214e-09,
+      "loss": 0.7396,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0610079575596816e-08,
+      "loss": 0.7346,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1936339522546418e-08,
+      "loss": 0.7206,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3262599469496022e-08,
+      "loss": 0.7395,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4588859416445623e-08,
+      "loss": 0.6768,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5915119363395225e-08,
+      "loss": 0.7811,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7241379310344825e-08,
+      "loss": 0.6933,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.856763925729443e-08,
+      "loss": 0.7514,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.989389920424403e-08,
+      "loss": 0.7314,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1220159151193633e-08,
+      "loss": 0.7244,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2546419098143233e-08,
+      "loss": 0.7248,
+      "step": 170
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3872679045092836e-08,
+      "loss": 0.7879,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.519893899204244e-08,
+      "loss": 0.7153,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6525198938992043e-08,
+      "loss": 0.668,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7851458885941643e-08,
+      "loss": 0.7278,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9177718832891247e-08,
+      "loss": 0.7566,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.050397877984085e-08,
+      "loss": 0.7281,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.183023872679045e-08,
+      "loss": 0.7775,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3156498673740054e-08,
+      "loss": 0.7088,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.448275862068965e-08,
+      "loss": 0.7009,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.580901856763926e-08,
+      "loss": 0.6174,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.713527851458886e-08,
+      "loss": 0.6453,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.846153846153846e-08,
+      "loss": 0.7588,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.978779840848806e-08,
+      "loss": 0.6935,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.111405835543767e-08,
+      "loss": 0.6947,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2440318302387265e-08,
+      "loss": 0.7923,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.376657824933687e-08,
+      "loss": 0.7317,
+      "step": 330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5092838196286466e-08,
+      "loss": 0.7249,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6419098143236076e-08,
+      "loss": 0.6902,
+      "step": 350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.774535809018567e-08,
+      "loss": 0.6815,
+      "step": 360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9071618037135276e-08,
+      "loss": 0.7027,
+      "step": 370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.039787798408488e-08,
+      "loss": 0.6319,
+      "step": 380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.172413793103448e-08,
+      "loss": 0.614,
+      "step": 390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.3050397877984086e-08,
+      "loss": 0.6328,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.437665782493368e-08,
+      "loss": 0.6642,
+      "step": 410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.570291777188329e-08,
+      "loss": 0.6395,
+      "step": 420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.7029177718832884e-08,
+      "loss": 0.6932,
+      "step": 430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.8355437665782494e-08,
+      "loss": 0.7672,
+      "step": 440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.968169761273208e-08,
+      "loss": 0.7185,
+      "step": 450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.10079575596817e-08,
+      "loss": 0.5958,
+      "step": 460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.233421750663129e-08,
+      "loss": 0.7257,
+      "step": 470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.36604774535809e-08,
+      "loss": 0.6535,
+      "step": 480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.49867374005305e-08,
+      "loss": 0.6412,
+      "step": 490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.631299734748011e-08,
+      "loss": 0.702,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.76392572944297e-08,
+      "loss": 0.6853,
+      "step": 510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.89655172413793e-08,
+      "loss": 0.6964,
+      "step": 520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.02917771883289e-08,
+      "loss": 0.6348,
+      "step": 530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.161803713527852e-08,
+      "loss": 0.6468,
+      "step": 540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.294429708222812e-08,
+      "loss": 0.684,
+      "step": 550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.427055702917772e-08,
+      "loss": 0.6557,
+      "step": 560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.559681697612731e-08,
+      "loss": 0.6512,
+      "step": 570
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.692307692307692e-08,
+      "loss": 0.6455,
+      "step": 580
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.824933687002652e-08,
+      "loss": 0.6924,
+      "step": 590
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.957559681697612e-08,
+      "loss": 0.6514,
+      "step": 600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.090185676392573e-08,
+      "loss": 0.6293,
+      "step": 610
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.222811671087534e-08,
+      "loss": 0.6318,
+      "step": 620
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.355437665782493e-08,
+      "loss": 0.6423,
+      "step": 630
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.488063660477453e-08,
+      "loss": 0.623,
+      "step": 640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.620689655172414e-08,
+      "loss": 0.6195,
+      "step": 650
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.753315649867374e-08,
+      "loss": 0.6402,
+      "step": 660
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.885941644562333e-08,
+      "loss": 0.6457,
+      "step": 670
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.018567639257293e-08,
+      "loss": 0.6561,
+      "step": 680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.151193633952254e-08,
+      "loss": 0.5819,
+      "step": 690
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.283819628647215e-08,
+      "loss": 0.6188,
+      "step": 700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.416445623342175e-08,
+      "loss": 0.6045,
+      "step": 710
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.549071618037134e-08,
+      "loss": 0.577,
+      "step": 720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.681697612732096e-08,
+      "loss": 0.6367,
+      "step": 730
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.814323607427055e-08,
+      "loss": 0.5688,
+      "step": 740
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.946949602122015e-08,
+      "loss": 0.5551,
+      "step": 750
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0079575596816976e-07,
+      "loss": 0.6775,
+      "step": 760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0212201591511936e-07,
+      "loss": 0.6046,
+      "step": 770
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0344827586206897e-07,
+      "loss": 0.7003,
+      "step": 780
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0477453580901856e-07,
+      "loss": 0.6015,
+      "step": 790
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0610079575596817e-07,
+      "loss": 0.6011,
+      "step": 800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0742705570291777e-07,
+      "loss": 0.5661,
+      "step": 810
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0875331564986737e-07,
+      "loss": 0.611,
+      "step": 820
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1007957559681696e-07,
+      "loss": 0.6089,
+      "step": 830
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1140583554376657e-07,
+      "loss": 0.6551,
+      "step": 840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1273209549071617e-07,
+      "loss": 0.575,
+      "step": 850
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1405835543766577e-07,
+      "loss": 0.5973,
+      "step": 860
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1538461538461539e-07,
+      "loss": 0.6034,
+      "step": 870
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1671087533156499e-07,
+      "loss": 0.6218,
+      "step": 880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1803713527851458e-07,
+      "loss": 0.5884,
+      "step": 890
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1936339522546417e-07,
+      "loss": 0.5975,
+      "step": 900
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.206896551724138e-07,
+      "loss": 0.6304,
+      "step": 910
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.220159151193634e-07,
+      "loss": 0.5389,
+      "step": 920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2334217506631299e-07,
+      "loss": 0.533,
+      "step": 930
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2466843501326258e-07,
+      "loss": 0.5926,
+      "step": 940
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.259946949602122e-07,
+      "loss": 0.5586,
+      "step": 950
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.273209549071618e-07,
+      "loss": 0.6088,
+      "step": 960
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.286472148541114e-07,
+      "loss": 0.6577,
+      "step": 970
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.29973474801061e-07,
+      "loss": 0.6132,
+      "step": 980
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.312997347480106e-07,
+      "loss": 0.5775,
+      "step": 990
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3262599469496022e-07,
+      "loss": 0.6712,
+      "step": 1000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.339522546419098e-07,
+      "loss": 0.6921,
+      "step": 1010
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.352785145888594e-07,
+      "loss": 0.6294,
+      "step": 1020
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.36604774535809e-07,
+      "loss": 0.5654,
+      "step": 1030
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.379310344827586e-07,
+      "loss": 0.5504,
+      "step": 1040
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.392572944297082e-07,
+      "loss": 0.619,
+      "step": 1050
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.405835543766578e-07,
+      "loss": 0.5249,
+      "step": 1060
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4190981432360742e-07,
+      "loss": 0.6055,
+      "step": 1070
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4323607427055704e-07,
+      "loss": 0.6498,
+      "step": 1080
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4456233421750664e-07,
+      "loss": 0.5519,
+      "step": 1090
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4588859416445624e-07,
+      "loss": 0.641,
+      "step": 1100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4721485411140583e-07,
+      "loss": 0.6033,
+      "step": 1110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4854111405835543e-07,
+      "loss": 0.5755,
+      "step": 1120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4986737400530503e-07,
+      "loss": 0.5373,
+      "step": 1130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5119363395225463e-07,
+      "loss": 0.4984,
+      "step": 1140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5251989389920425e-07,
+      "loss": 0.6086,
+      "step": 1150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5384615384615385e-07,
+      "loss": 0.4949,
+      "step": 1160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5517241379310344e-07,
+      "loss": 0.525,
+      "step": 1170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5649867374005304e-07,
+      "loss": 0.5949,
+      "step": 1180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5782493368700264e-07,
+      "loss": 0.5532,
+      "step": 1190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5915119363395223e-07,
+      "loss": 0.4981,
+      "step": 1200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6047745358090183e-07,
+      "loss": 0.6167,
+      "step": 1210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6180371352785145e-07,
+      "loss": 0.545,
+      "step": 1220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6312997347480108e-07,
+      "loss": 0.5325,
+      "step": 1230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6445623342175067e-07,
+      "loss": 0.5609,
+      "step": 1240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6578249336870027e-07,
+      "loss": 0.606,
+      "step": 1250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6710875331564987e-07,
+      "loss": 0.5249,
+      "step": 1260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6843501326259946e-07,
+      "loss": 0.3742,
+      "step": 1270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6976127320954906e-07,
+      "loss": 0.5403,
+      "step": 1280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7108753315649866e-07,
+      "loss": 0.5028,
+      "step": 1290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7241379310344828e-07,
+      "loss": 0.5407,
+      "step": 1300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7374005305039788e-07,
+      "loss": 0.5825,
+      "step": 1310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7506631299734747e-07,
+      "loss": 0.5605,
+      "step": 1320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7639257294429707e-07,
+      "loss": 0.4911,
+      "step": 1330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7771883289124667e-07,
+      "loss": 0.4465,
+      "step": 1340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7904509283819627e-07,
+      "loss": 0.5397,
+      "step": 1350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8037135278514586e-07,
+      "loss": 0.4645,
+      "step": 1360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8169761273209549e-07,
+      "loss": 0.569,
+      "step": 1370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8302387267904508e-07,
+      "loss": 0.4948,
+      "step": 1380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8435013262599468e-07,
+      "loss": 0.6482,
+      "step": 1390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.856763925729443e-07,
+      "loss": 0.5004,
+      "step": 1400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.870026525198939e-07,
+      "loss": 0.5158,
+      "step": 1410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.883289124668435e-07,
+      "loss": 0.5585,
+      "step": 1420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.896551724137931e-07,
+      "loss": 0.4894,
+      "step": 1430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.909814323607427e-07,
+      "loss": 0.5977,
+      "step": 1440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9230769230769231e-07,
+      "loss": 0.4575,
+      "step": 1450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.936339522546419e-07,
+      "loss": 0.5371,
+      "step": 1460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.949602122015915e-07,
+      "loss": 0.6611,
+      "step": 1470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.962864721485411e-07,
+      "loss": 0.4122,
+      "step": 1480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.976127320954907e-07,
+      "loss": 0.4914,
+      "step": 1490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.989389920424403e-07,
+      "loss": 0.535,
+      "step": 1500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.002652519893899e-07,
+      "loss": 0.5063,
+      "step": 1510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0159151193633952e-07,
+      "loss": 0.5164,
+      "step": 1520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0291777188328911e-07,
+      "loss": 0.5227,
+      "step": 1530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.042440318302387e-07,
+      "loss": 0.4943,
+      "step": 1540
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.055702917771883e-07,
+      "loss": 0.4948,
+      "step": 1550
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0689655172413793e-07,
+      "loss": 0.5938,
+      "step": 1560
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0822281167108753e-07,
+      "loss": 0.6162,
+      "step": 1570
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0954907161803713e-07,
+      "loss": 0.5321,
+      "step": 1580
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1087533156498675e-07,
+      "loss": 0.6596,
+      "step": 1590
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1220159151193635e-07,
+      "loss": 0.5412,
+      "step": 1600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1352785145888594e-07,
+      "loss": 0.6187,
+      "step": 1610
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1485411140583554e-07,
+      "loss": 0.5271,
+      "step": 1620
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1618037135278514e-07,
+      "loss": 0.6625,
+      "step": 1630
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1750663129973473e-07,
+      "loss": 0.5927,
+      "step": 1640
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1883289124668433e-07,
+      "loss": 0.6209,
+      "step": 1650
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.2015915119363393e-07,
+      "loss": 0.5249,
+      "step": 1660
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.2148541114058355e-07,
+      "loss": 0.5256,
+      "step": 1670
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.2281167108753315e-07,
+      "loss": 0.4763,
+      "step": 1680
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.2413793103448274e-07,
+      "loss": 0.4851,
+      "step": 1690
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.2546419098143234e-07,
+      "loss": 0.4954,
+      "step": 1700
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.2679045092838194e-07,
+      "loss": 0.4942,
+      "step": 1710
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.2811671087533153e-07,
+      "loss": 0.415,
+      "step": 1720
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.2944297082228116e-07,
+      "loss": 0.5134,
+      "step": 1730
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3076923076923078e-07,
+      "loss": 0.7008,
+      "step": 1740
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3209549071618038e-07,
+      "loss": 0.5747,
+      "step": 1750
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3342175066312997e-07,
+      "loss": 0.4681,
+      "step": 1760
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3474801061007957e-07,
+      "loss": 0.4727,
+      "step": 1770
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3607427055702917e-07,
+      "loss": 0.5353,
+      "step": 1780
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3740053050397877e-07,
+      "loss": 0.6287,
+      "step": 1790
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3872679045092834e-07,
+      "loss": 0.4294,
+      "step": 1800
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.40053050397878e-07,
+      "loss": 0.5166,
+      "step": 1810
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.413793103448276e-07,
+      "loss": 0.4971,
+      "step": 1820
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.427055702917772e-07,
+      "loss": 0.4158,
+      "step": 1830
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.440318302387268e-07,
+      "loss": 0.4804,
+      "step": 1840
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4535809018567637e-07,
+      "loss": 0.4916,
+      "step": 1850
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4668435013262597e-07,
+      "loss": 0.6046,
+      "step": 1860
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4801061007957557e-07,
+      "loss": 0.5051,
+      "step": 1870
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4933687002652516e-07,
+      "loss": 0.6202,
+      "step": 1880
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5066312997347476e-07,
+      "loss": 0.5563,
+      "step": 1890
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.519893899204244e-07,
+      "loss": 0.5172,
+      "step": 1900
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5331564986737395e-07,
+      "loss": 0.4743,
+      "step": 1910
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.546419098143236e-07,
+      "loss": 0.5185,
+      "step": 1920
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.559681697612732e-07,
+      "loss": 0.5641,
+      "step": 1930
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.572944297082228e-07,
+      "loss": 0.446,
+      "step": 1940
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.586206896551724e-07,
+      "loss": 0.5778,
+      "step": 1950
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.59946949602122e-07,
+      "loss": 0.5114,
+      "step": 1960
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6127320954907164e-07,
+      "loss": 0.5077,
+      "step": 1970
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.625994694960212e-07,
+      "loss": 0.5656,
+      "step": 1980
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6392572944297084e-07,
+      "loss": 0.4638,
+      "step": 1990
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6525198938992043e-07,
+      "loss": 0.5302,
+      "step": 2000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6657824933687003e-07,
+      "loss": 0.53,
+      "step": 2010
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.679045092838196e-07,
+      "loss": 0.3901,
+      "step": 2020
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.692307692307692e-07,
+      "loss": 0.4339,
+      "step": 2030
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.705570291777188e-07,
+      "loss": 0.4862,
+      "step": 2040
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.718832891246684e-07,
+      "loss": 0.4803,
+      "step": 2050
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.73209549071618e-07,
+      "loss": 0.4935,
+      "step": 2060
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.7453580901856766e-07,
+      "loss": 0.5879,
+      "step": 2070
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.758620689655172e-07,
+      "loss": 0.5677,
+      "step": 2080
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.7718832891246686e-07,
+      "loss": 0.5324,
+      "step": 2090
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.785145888594164e-07,
+      "loss": 0.3662,
+      "step": 2100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.7984084880636605e-07,
+      "loss": 0.4914,
+      "step": 2110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.811671087533156e-07,
+      "loss": 0.5198,
+      "step": 2120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8249336870026524e-07,
+      "loss": 0.4428,
+      "step": 2130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8381962864721484e-07,
+      "loss": 0.4997,
+      "step": 2140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8514588859416444e-07,
+      "loss": 0.5166,
+      "step": 2150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.864721485411141e-07,
+      "loss": 0.5251,
+      "step": 2160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8779840848806363e-07,
+      "loss": 0.5525,
+      "step": 2170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.891246684350133e-07,
+      "loss": 0.4615,
+      "step": 2180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.904509283819628e-07,
+      "loss": 0.5127,
+      "step": 2190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.917771883289125e-07,
+      "loss": 0.5429,
+      "step": 2200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.93103448275862e-07,
+      "loss": 0.4352,
+      "step": 2210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9442970822281167e-07,
+      "loss": 0.4291,
+      "step": 2220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9575596816976127e-07,
+      "loss": 0.4928,
+      "step": 2230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9708222811671086e-07,
+      "loss": 0.4417,
+      "step": 2240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9840848806366046e-07,
+      "loss": 0.4848,
+      "step": 2250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9973474801061006e-07,
+      "loss": 0.4583,
+      "step": 2260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.0106100795755965e-07,
+      "loss": 0.6345,
+      "step": 2270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.0238726790450925e-07,
+      "loss": 0.4416,
+      "step": 2280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.0371352785145885e-07,
+      "loss": 0.4679,
+      "step": 2290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.050397877984085e-07,
+      "loss": 0.6144,
+      "step": 2300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.063660477453581e-07,
+      "loss": 0.5629,
+      "step": 2310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.076923076923077e-07,
+      "loss": 0.6378,
+      "step": 2320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.090185676392573e-07,
+      "loss": 0.4866,
+      "step": 2330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.103448275862069e-07,
+      "loss": 0.431,
+      "step": 2340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.116710875331565e-07,
+      "loss": 0.5033,
+      "step": 2350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.129973474801061e-07,
+      "loss": 0.613,
+      "step": 2360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.1432360742705573e-07,
+      "loss": 0.4152,
+      "step": 2370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.1564986737400527e-07,
+      "loss": 0.5032,
+      "step": 2380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.169761273209549e-07,
+      "loss": 0.5202,
+      "step": 2390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.1830238726790447e-07,
+      "loss": 0.4472,
+      "step": 2400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.196286472148541e-07,
+      "loss": 0.5215,
+      "step": 2410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.2095490716180366e-07,
+      "loss": 0.5738,
+      "step": 2420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.222811671087533e-07,
+      "loss": 0.481,
+      "step": 2430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.236074270557029e-07,
+      "loss": 0.4373,
+      "step": 2440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.249336870026525e-07,
+      "loss": 0.5769,
+      "step": 2450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2625994694960215e-07,
+      "loss": 0.5123,
+      "step": 2460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.275862068965517e-07,
+      "loss": 0.4273,
+      "step": 2470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2891246684350135e-07,
+      "loss": 0.6897,
+      "step": 2480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.302387267904509e-07,
+      "loss": 0.5386,
+      "step": 2490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3156498673740054e-07,
+      "loss": 0.4524,
+      "step": 2500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.328912466843501e-07,
+      "loss": 0.4212,
+      "step": 2510
+    },
+    {
+      "epoch": 0.07,
+      "eval_accuracy": 0.7236377891023128,
+      "eval_loss": 0.4736761152744293,
+      "eval_runtime": 719.9771,
+      "eval_samples_per_second": 3.543,
+      "eval_steps_per_second": 0.886,
+      "step": 2514
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3421750663129973e-07,
+      "loss": 0.45,
+      "step": 2520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3554376657824933e-07,
+      "loss": 0.5435,
+      "step": 2530
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3687002652519893e-07,
+      "loss": 0.5493,
+      "step": 2540
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.381962864721485e-07,
+      "loss": 0.4906,
+      "step": 2550
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.395225464190981e-07,
+      "loss": 0.425,
+      "step": 2560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.408488063660477e-07,
+      "loss": 0.4625,
+      "step": 2570
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.421750663129973e-07,
+      "loss": 0.4946,
+      "step": 2580
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.435013262599469e-07,
+      "loss": 0.527,
+      "step": 2590
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.4482758620689656e-07,
+      "loss": 0.4673,
+      "step": 2600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.461538461538461e-07,
+      "loss": 0.4973,
+      "step": 2610
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.4748010610079576e-07,
+      "loss": 0.5398,
+      "step": 2620
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.4880636604774535e-07,
+      "loss": 0.5386,
+      "step": 2630
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.5013262599469495e-07,
+      "loss": 0.4936,
+      "step": 2640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.5145888594164455e-07,
+      "loss": 0.4297,
+      "step": 2650
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.5278514588859414e-07,
+      "loss": 0.6891,
+      "step": 2660
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.541114058355438e-07,
+      "loss": 0.3811,
+      "step": 2670
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.5543766578249334e-07,
+      "loss": 0.5439,
+      "step": 2680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.56763925729443e-07,
+      "loss": 0.4031,
+      "step": 2690
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.5809018567639253e-07,
+      "loss": 0.6409,
+      "step": 2700
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.594164456233422e-07,
+      "loss": 0.3939,
+      "step": 2710
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.607427055702917e-07,
+      "loss": 0.5025,
+      "step": 2720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.620689655172414e-07,
+      "loss": 0.3268,
+      "step": 2730
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.6339522546419097e-07,
+      "loss": 0.3455,
+      "step": 2740
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.6472148541114057e-07,
+      "loss": 0.4999,
+      "step": 2750
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.6604774535809016e-07,
+      "loss": 0.4002,
+      "step": 2760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.6737400530503976e-07,
+      "loss": 0.4158,
+      "step": 2770
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.6870026525198936e-07,
+      "loss": 0.5939,
+      "step": 2780
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7002652519893896e-07,
+      "loss": 0.3991,
+      "step": 2790
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.713527851458886e-07,
+      "loss": 0.3546,
+      "step": 2800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.726790450928382e-07,
+      "loss": 0.3139,
+      "step": 2810
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.740053050397878e-07,
+      "loss": 0.5089,
+      "step": 2820
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.753315649867374e-07,
+      "loss": 0.5739,
+      "step": 2830
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.76657824933687e-07,
+      "loss": 0.6196,
+      "step": 2840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.779840848806366e-07,
+      "loss": 0.5625,
+      "step": 2850
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.793103448275862e-07,
+      "loss": 0.6529,
+      "step": 2860
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.806366047745358e-07,
+      "loss": 0.393,
+      "step": 2870
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.819628647214854e-07,
+      "loss": 0.4224,
+      "step": 2880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.83289124668435e-07,
+      "loss": 0.3901,
+      "step": 2890
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8461538461538463e-07,
+      "loss": 0.4191,
+      "step": 2900
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8594164456233417e-07,
+      "loss": 0.4255,
+      "step": 2910
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.872679045092838e-07,
+      "loss": 0.5912,
+      "step": 2920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8859416445623336e-07,
+      "loss": 0.5561,
+      "step": 2930
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.89920424403183e-07,
+      "loss": 0.5536,
+      "step": 2940
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9124668435013256e-07,
+      "loss": 0.3742,
+      "step": 2950
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.925729442970822e-07,
+      "loss": 0.4141,
+      "step": 2960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9389920424403186e-07,
+      "loss": 0.5263,
+      "step": 2970
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.952254641909814e-07,
+      "loss": 0.4386,
+      "step": 2980
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9655172413793105e-07,
+      "loss": 0.494,
+      "step": 2990
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.978779840848806e-07,
+      "loss": 0.4584,
+      "step": 3000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9920424403183024e-07,
+      "loss": 0.5082,
+      "step": 3010
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.005305039787798e-07,
+      "loss": 0.2946,
+      "step": 3020
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0185676392572944e-07,
+      "loss": 0.4121,
+      "step": 3030
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0318302387267904e-07,
+      "loss": 0.4016,
+      "step": 3040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0450928381962863e-07,
+      "loss": 0.5245,
+      "step": 3050
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0583554376657823e-07,
+      "loss": 0.4977,
+      "step": 3060
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.071618037135278e-07,
+      "loss": 0.4575,
+      "step": 3070
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.084880636604774e-07,
+      "loss": 0.5884,
+      "step": 3080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.09814323607427e-07,
+      "loss": 0.4351,
+      "step": 3090
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.111405835543766e-07,
+      "loss": 0.4785,
+      "step": 3100
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1246684350132627e-07,
+      "loss": 0.4365,
+      "step": 3110
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1379310344827586e-07,
+      "loss": 0.5421,
+      "step": 3120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1511936339522546e-07,
+      "loss": 0.4801,
+      "step": 3130
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1644562334217506e-07,
+      "loss": 0.4038,
+      "step": 3140
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1777188328912465e-07,
+      "loss": 0.3979,
+      "step": 3150
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1909814323607425e-07,
+      "loss": 0.4959,
+      "step": 3160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2042440318302385e-07,
+      "loss": 0.5416,
+      "step": 3170
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.217506631299735e-07,
+      "loss": 0.5041,
+      "step": 3180
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2307692307692304e-07,
+      "loss": 0.4169,
+      "step": 3190
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.244031830238727e-07,
+      "loss": 0.3745,
+      "step": 3200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2572944297082224e-07,
+      "loss": 0.3346,
+      "step": 3210
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.270557029177719e-07,
+      "loss": 0.556,
+      "step": 3220
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2838196286472143e-07,
+      "loss": 0.4442,
+      "step": 3230
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.297082228116711e-07,
+      "loss": 0.3385,
+      "step": 3240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.310344827586206e-07,
+      "loss": 0.4279,
+      "step": 3250
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.3236074270557027e-07,
+      "loss": 0.4956,
+      "step": 3260
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.3368700265251987e-07,
+      "loss": 0.5013,
+      "step": 3270
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.3501326259946947e-07,
+      "loss": 0.3458,
+      "step": 3280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.363395225464191e-07,
+      "loss": 0.3741,
+      "step": 3290
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.3766578249336866e-07,
+      "loss": 0.4561,
+      "step": 3300
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.389920424403183e-07,
+      "loss": 0.2777,
+      "step": 3310
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4031830238726785e-07,
+      "loss": 0.4734,
+      "step": 3320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.416445623342175e-07,
+      "loss": 0.4401,
+      "step": 3330
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.429708222811671e-07,
+      "loss": 0.4014,
+      "step": 3340
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.442970822281167e-07,
+      "loss": 0.5563,
+      "step": 3350
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.456233421750663e-07,
+      "loss": 0.5158,
+      "step": 3360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.469496021220159e-07,
+      "loss": 0.2965,
+      "step": 3370
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.482758620689655e-07,
+      "loss": 0.4342,
+      "step": 3380
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.496021220159151e-07,
+      "loss": 0.6739,
+      "step": 3390
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.509283819628647e-07,
+      "loss": 0.4688,
+      "step": 3400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5225464190981433e-07,
+      "loss": 0.6134,
+      "step": 3410
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.535809018567639e-07,
+      "loss": 0.5544,
+      "step": 3420
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.549071618037135e-07,
+      "loss": 0.3696,
+      "step": 3430
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5623342175066307e-07,
+      "loss": 0.4234,
+      "step": 3440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.575596816976127e-07,
+      "loss": 0.415,
+      "step": 3450
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.588859416445623e-07,
+      "loss": 0.3957,
+      "step": 3460
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.602122015915119e-07,
+      "loss": 0.4459,
+      "step": 3470
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6153846153846156e-07,
+      "loss": 0.4735,
+      "step": 3480
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.628647214854111e-07,
+      "loss": 0.3834,
+      "step": 3490
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6419098143236076e-07,
+      "loss": 0.3422,
+      "step": 3500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.655172413793103e-07,
+      "loss": 0.3794,
+      "step": 3510
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6684350132625995e-07,
+      "loss": 0.511,
+      "step": 3520
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.681697612732095e-07,
+      "loss": 0.5311,
+      "step": 3530
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6949602122015914e-07,
+      "loss": 0.6242,
+      "step": 3540
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.7082228116710874e-07,
+      "loss": 0.2774,
+      "step": 3550
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.7214854111405834e-07,
+      "loss": 0.41,
+      "step": 3560
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.7347480106100793e-07,
+      "loss": 0.4245,
+      "step": 3570
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.7480106100795753e-07,
+      "loss": 0.321,
+      "step": 3580
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.7612732095490713e-07,
+      "loss": 0.3891,
+      "step": 3590
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.774535809018567e-07,
+      "loss": 0.4696,
+      "step": 3600
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.787798408488063e-07,
+      "loss": 0.5748,
+      "step": 3610
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.80106100795756e-07,
+      "loss": 0.4338,
+      "step": 3620
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.814323607427055e-07,
+      "loss": 0.3506,
+      "step": 3630
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.827586206896552e-07,
+      "loss": 0.4195,
+      "step": 3640
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.840848806366047e-07,
+      "loss": 0.3021,
+      "step": 3650
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.854111405835544e-07,
+      "loss": 0.3837,
+      "step": 3660
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.867374005305039e-07,
+      "loss": 0.5144,
+      "step": 3670
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.880636604774536e-07,
+      "loss": 0.331,
+      "step": 3680
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.893899204244031e-07,
+      "loss": 0.5365,
+      "step": 3690
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.907161803713527e-07,
+      "loss": 0.5182,
+      "step": 3700
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.920424403183024e-07,
+      "loss": 0.3856,
+      "step": 3710
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.933687002652519e-07,
+      "loss": 0.3803,
+      "step": 3720
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.946949602122016e-07,
+      "loss": 0.3418,
+      "step": 3730
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.960212201591511e-07,
+      "loss": 0.4443,
+      "step": 3740
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.973474801061008e-07,
+      "loss": 0.4819,
+      "step": 3750
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.986737400530503e-07,
+      "loss": 0.3945,
+      "step": 3760
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5e-07,
+      "loss": 0.5144,
+      "step": 3770
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.998526160648489e-07,
+      "loss": 0.3183,
+      "step": 3780
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.997052321296979e-07,
+      "loss": 0.4734,
+      "step": 3790
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.995578481945468e-07,
+      "loss": 0.4157,
+      "step": 3800
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.994104642593957e-07,
+      "loss": 0.5323,
+      "step": 3810
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.992630803242447e-07,
+      "loss": 0.4605,
+      "step": 3820
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.991156963890936e-07,
+      "loss": 0.3047,
+      "step": 3830
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.989683124539425e-07,
+      "loss": 0.3728,
+      "step": 3840
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.988209285187915e-07,
+      "loss": 0.2399,
+      "step": 3850
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.986735445836404e-07,
+      "loss": 0.5681,
+      "step": 3860
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.985261606484893e-07,
+      "loss": 0.3124,
+      "step": 3870
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.983787767133383e-07,
+      "loss": 0.3382,
+      "step": 3880
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.982313927781872e-07,
+      "loss": 0.2701,
+      "step": 3890
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.980840088430361e-07,
+      "loss": 0.2886,
+      "step": 3900
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.97936624907885e-07,
+      "loss": 0.2763,
+      "step": 3910
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.97789240972734e-07,
+      "loss": 0.4194,
+      "step": 3920
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.976418570375829e-07,
+      "loss": 0.6321,
+      "step": 3930
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.974944731024318e-07,
+      "loss": 0.2999,
+      "step": 3940
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.973470891672808e-07,
+      "loss": 0.4275,
+      "step": 3950
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.971997052321297e-07,
+      "loss": 0.3593,
+      "step": 3960
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.970523212969786e-07,
+      "loss": 0.4066,
+      "step": 3970
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.969049373618276e-07,
+      "loss": 0.3789,
+      "step": 3980
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.967575534266765e-07,
+      "loss": 0.2811,
+      "step": 3990
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.966101694915254e-07,
+      "loss": 0.4003,
+      "step": 4000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.964627855563744e-07,
+      "loss": 0.3358,
+      "step": 4010
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.963154016212233e-07,
+      "loss": 0.4802,
+      "step": 4020
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.961680176860722e-07,
+      "loss": 0.2442,
+      "step": 4030
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.960206337509212e-07,
+      "loss": 0.4891,
+      "step": 4040
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.958732498157701e-07,
+      "loss": 0.2671,
+      "step": 4050
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.95725865880619e-07,
+      "loss": 0.3811,
+      "step": 4060
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.95578481945468e-07,
+      "loss": 0.3919,
+      "step": 4070
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.954310980103169e-07,
+      "loss": 0.506,
+      "step": 4080
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.952837140751658e-07,
+      "loss": 0.2994,
+      "step": 4090
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.951363301400148e-07,
+      "loss": 0.3867,
+      "step": 4100
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.949889462048637e-07,
+      "loss": 0.3273,
+      "step": 4110
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.948415622697126e-07,
+      "loss": 0.4552,
+      "step": 4120
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.946941783345616e-07,
+      "loss": 0.4559,
+      "step": 4130
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.945467943994105e-07,
+      "loss": 0.4017,
+      "step": 4140
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.943994104642594e-07,
+      "loss": 0.324,
+      "step": 4150
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.942520265291082e-07,
+      "loss": 0.4674,
+      "step": 4160
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.941046425939573e-07,
+      "loss": 0.418,
+      "step": 4170
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.939572586588061e-07,
+      "loss": 0.2469,
+      "step": 4180
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.93809874723655e-07,
+      "loss": 0.4169,
+      "step": 4190
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.93662490788504e-07,
+      "loss": 0.514,
+      "step": 4200
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.93515106853353e-07,
+      "loss": 0.2565,
+      "step": 4210
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.933677229182018e-07,
+      "loss": 0.2596,
+      "step": 4220
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.932203389830508e-07,
+      "loss": 0.3401,
+      "step": 4230
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.930729550478997e-07,
+      "loss": 0.3624,
+      "step": 4240
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.929255711127486e-07,
+      "loss": 0.406,
+      "step": 4250
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.927781871775976e-07,
+      "loss": 0.4248,
+      "step": 4260
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.926308032424465e-07,
+      "loss": 0.3617,
+      "step": 4270
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.924834193072954e-07,
+      "loss": 0.358,
+      "step": 4280
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.923360353721444e-07,
+      "loss": 0.3392,
+      "step": 4290
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.921886514369933e-07,
+      "loss": 0.311,
+      "step": 4300
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.920412675018422e-07,
+      "loss": 0.3526,
+      "step": 4310
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.918938835666912e-07,
+      "loss": 0.3639,
+      "step": 4320
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.917464996315401e-07,
+      "loss": 0.4344,
+      "step": 4330
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.91599115696389e-07,
+      "loss": 0.4051,
+      "step": 4340
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.91451731761238e-07,
+      "loss": 0.32,
+      "step": 4350
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.913043478260869e-07,
+      "loss": 0.2914,
+      "step": 4360
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.911569638909358e-07,
+      "loss": 0.1726,
+      "step": 4370
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.910095799557848e-07,
+      "loss": 0.4132,
+      "step": 4380
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.908621960206337e-07,
+      "loss": 0.3157,
+      "step": 4390
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.907148120854826e-07,
+      "loss": 0.2615,
+      "step": 4400
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.905674281503316e-07,
+      "loss": 0.3865,
+      "step": 4410
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.904200442151805e-07,
+      "loss": 0.2383,
+      "step": 4420
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.902726602800294e-07,
+      "loss": 0.5915,
+      "step": 4430
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.901252763448783e-07,
+      "loss": 0.2984,
+      "step": 4440
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.899778924097273e-07,
+      "loss": 0.2982,
+      "step": 4450
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.898305084745762e-07,
+      "loss": 0.3448,
+      "step": 4460
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.896831245394251e-07,
+      "loss": 0.3028,
+      "step": 4470
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.895357406042741e-07,
+      "loss": 0.5457,
+      "step": 4480
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.89388356669123e-07,
+      "loss": 0.2775,
+      "step": 4490
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.892409727339719e-07,
+      "loss": 0.3331,
+      "step": 4500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.890935887988209e-07,
+      "loss": 0.2089,
+      "step": 4510
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.889462048636698e-07,
+      "loss": 0.4544,
+      "step": 4520
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.887988209285187e-07,
+      "loss": 0.1698,
+      "step": 4530
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.886514369933677e-07,
+      "loss": 0.4611,
+      "step": 4540
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.885040530582166e-07,
+      "loss": 0.3863,
+      "step": 4550
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.883566691230655e-07,
+      "loss": 0.3619,
+      "step": 4560
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.882092851879145e-07,
+      "loss": 0.2803,
+      "step": 4570
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.880619012527634e-07,
+      "loss": 0.3004,
+      "step": 4580
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.879145173176123e-07,
+      "loss": 0.4303,
+      "step": 4590
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.877671333824613e-07,
+      "loss": 0.3028,
+      "step": 4600
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.876197494473102e-07,
+      "loss": 0.3317,
+      "step": 4610
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.874723655121591e-07,
+      "loss": 0.656,
+      "step": 4620
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.873249815770081e-07,
+      "loss": 0.2668,
+      "step": 4630
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.87177597641857e-07,
+      "loss": 0.2993,
+      "step": 4640
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.870302137067059e-07,
+      "loss": 0.3879,
+      "step": 4650
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.868828297715549e-07,
+      "loss": 0.2357,
+      "step": 4660
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.867354458364038e-07,
+      "loss": 0.4819,
+      "step": 4670
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.865880619012527e-07,
+      "loss": 0.3588,
+      "step": 4680
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.864406779661016e-07,
+      "loss": 0.2651,
+      "step": 4690
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.862932940309506e-07,
+      "loss": 0.3486,
+      "step": 4700
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.861459100957995e-07,
+      "loss": 0.2048,
+      "step": 4710
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.859985261606484e-07,
+      "loss": 0.2247,
+      "step": 4720
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.858511422254974e-07,
+      "loss": 0.2861,
+      "step": 4730
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.857037582903463e-07,
+      "loss": 0.2153,
+      "step": 4740
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.855563743551952e-07,
+      "loss": 0.2081,
+      "step": 4750
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.854089904200442e-07,
+      "loss": 0.448,
+      "step": 4760
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.852616064848931e-07,
+      "loss": 0.238,
+      "step": 4770
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.85114222549742e-07,
+      "loss": 0.1986,
+      "step": 4780
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.84966838614591e-07,
+      "loss": 0.1439,
+      "step": 4790
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.848194546794399e-07,
+      "loss": 0.441,
+      "step": 4800
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.846720707442888e-07,
+      "loss": 0.301,
+      "step": 4810
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.845246868091378e-07,
+      "loss": 0.5525,
+      "step": 4820
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.843773028739867e-07,
+      "loss": 0.2238,
+      "step": 4830
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.842299189388356e-07,
+      "loss": 0.2699,
+      "step": 4840
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.840825350036846e-07,
+      "loss": 0.2055,
+      "step": 4850
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.839351510685335e-07,
+      "loss": 0.3016,
+      "step": 4860
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.837877671333824e-07,
+      "loss": 0.3008,
+      "step": 4870
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.836403831982314e-07,
+      "loss": 0.3061,
+      "step": 4880
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.834929992630803e-07,
+      "loss": 0.2781,
+      "step": 4890
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.833456153279292e-07,
+      "loss": 0.2262,
+      "step": 4900
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.831982313927782e-07,
+      "loss": 0.6202,
+      "step": 4910
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.830508474576271e-07,
+      "loss": 0.2492,
+      "step": 4920
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.82903463522476e-07,
+      "loss": 0.2977,
+      "step": 4930
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.827560795873249e-07,
+      "loss": 0.3409,
+      "step": 4940
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.826086956521739e-07,
+      "loss": 0.4883,
+      "step": 4950
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.824613117170228e-07,
+      "loss": 0.3896,
+      "step": 4960
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.823139277818717e-07,
+      "loss": 0.4083,
+      "step": 4970
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.821665438467207e-07,
+      "loss": 0.3097,
+      "step": 4980
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.820191599115696e-07,
+      "loss": 0.2651,
+      "step": 4990
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.818717759764185e-07,
+      "loss": 0.129,
+      "step": 5000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.817243920412675e-07,
+      "loss": 0.2052,
+      "step": 5010
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.815770081061164e-07,
+      "loss": 0.2813,
+      "step": 5020
+    },
+    {
+      "epoch": 1.07,
+      "eval_accuracy": 0.8349666797334379,
+      "eval_loss": 0.3921673893928528,
+      "eval_runtime": 678.094,
+      "eval_samples_per_second": 3.762,
+      "eval_steps_per_second": 0.941,
+      "step": 5028
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.814296241709653e-07,
+      "loss": 0.4054,
+      "step": 5030
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.812822402358143e-07,
+      "loss": 0.3766,
+      "step": 5040
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.811348563006632e-07,
+      "loss": 0.1505,
+      "step": 5050
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.809874723655121e-07,
+      "loss": 0.53,
+      "step": 5060
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.808400884303611e-07,
+      "loss": 0.2959,
+      "step": 5070
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.8069270449521e-07,
+      "loss": 0.4601,
+      "step": 5080
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.805453205600589e-07,
+      "loss": 0.3165,
+      "step": 5090
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.803979366249079e-07,
+      "loss": 0.4869,
+      "step": 5100
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.802505526897568e-07,
+      "loss": 0.229,
+      "step": 5110
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.801031687546057e-07,
+      "loss": 0.4105,
+      "step": 5120
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.799557848194547e-07,
+      "loss": 0.2917,
+      "step": 5130
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.798084008843036e-07,
+      "loss": 0.2166,
+      "step": 5140
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.796610169491525e-07,
+      "loss": 0.2162,
+      "step": 5150
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.795136330140015e-07,
+      "loss": 0.3397,
+      "step": 5160
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.793662490788504e-07,
+      "loss": 0.2958,
+      "step": 5170
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.792188651436993e-07,
+      "loss": 0.3242,
+      "step": 5180
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.790714812085483e-07,
+      "loss": 0.671,
+      "step": 5190
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.789240972733972e-07,
+      "loss": 0.2393,
+      "step": 5200
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.787767133382461e-07,
+      "loss": 0.1399,
+      "step": 5210
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.78629329403095e-07,
+      "loss": 0.6508,
+      "step": 5220
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.78481945467944e-07,
+      "loss": 0.6255,
+      "step": 5230
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.783345615327929e-07,
+      "loss": 0.4356,
+      "step": 5240
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.781871775976418e-07,
+      "loss": 0.513,
+      "step": 5250
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.780397936624908e-07,
+      "loss": 0.2814,
+      "step": 5260
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.778924097273397e-07,
+      "loss": 0.2206,
+      "step": 5270
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.777450257921886e-07,
+      "loss": 0.1432,
+      "step": 5280
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.775976418570376e-07,
+      "loss": 0.1767,
+      "step": 5290
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.774502579218865e-07,
+      "loss": 0.2453,
+      "step": 5300
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.773028739867354e-07,
+      "loss": 0.2471,
+      "step": 5310
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.771554900515844e-07,
+      "loss": 0.2202,
+      "step": 5320
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.770081061164333e-07,
+      "loss": 0.2848,
+      "step": 5330
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.768607221812822e-07,
+      "loss": 0.1851,
+      "step": 5340
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.767133382461311e-07,
+      "loss": 0.2146,
+      "step": 5350
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.7656595431098007e-07,
+      "loss": 0.2224,
+      "step": 5360
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.76418570375829e-07,
+      "loss": 0.2458,
+      "step": 5370
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.762711864406779e-07,
+      "loss": 0.4894,
+      "step": 5380
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.7612380250552687e-07,
+      "loss": 0.3553,
+      "step": 5390
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.7597641857037577e-07,
+      "loss": 0.1784,
+      "step": 5400
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.758290346352247e-07,
+      "loss": 0.3073,
+      "step": 5410
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.7568165070007367e-07,
+      "loss": 0.2926,
+      "step": 5420
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.7553426676492257e-07,
+      "loss": 0.3976,
+      "step": 5430
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.753868828297715e-07,
+      "loss": 0.4535,
+      "step": 5440
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.7523949889462047e-07,
+      "loss": 0.1169,
+      "step": 5450
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.7509211495946936e-07,
+      "loss": 0.1739,
+      "step": 5460
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.749447310243183e-07,
+      "loss": 0.5978,
+      "step": 5470
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.7479734708916727e-07,
+      "loss": 0.4071,
+      "step": 5480
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.7464996315401616e-07,
+      "loss": 0.1884,
+      "step": 5490
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.745025792188651e-07,
+      "loss": 0.3179,
+      "step": 5500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.7435519528371406e-07,
+      "loss": 0.2914,
+      "step": 5510
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.7420781134856296e-07,
+      "loss": 0.4433,
+      "step": 5520
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.740604274134119e-07,
+      "loss": 0.3206,
+      "step": 5530
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.739130434782608e-07,
+      "loss": 0.4274,
+      "step": 5540
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.7376565954310976e-07,
+      "loss": 0.3956,
+      "step": 5550
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.736182756079587e-07,
+      "loss": 0.1281,
+      "step": 5560
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.734708916728076e-07,
+      "loss": 0.1874,
+      "step": 5570
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.7332350773765656e-07,
+      "loss": 0.1682,
+      "step": 5580
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.731761238025055e-07,
+      "loss": 0.2154,
+      "step": 5590
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.730287398673544e-07,
+      "loss": 0.3607,
+      "step": 5600
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.7288135593220336e-07,
+      "loss": 0.2851,
+      "step": 5610
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.727339719970523e-07,
+      "loss": 0.0615,
+      "step": 5620
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.725865880619012e-07,
+      "loss": 0.1584,
+      "step": 5630
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.7243920412675015e-07,
+      "loss": 0.4012,
+      "step": 5640
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.7229182019159905e-07,
+      "loss": 0.0663,
+      "step": 5650
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.72144436256448e-07,
+      "loss": 0.1379,
+      "step": 5660
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.7199705232129695e-07,
+      "loss": 0.1469,
+      "step": 5670
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.7184966838614585e-07,
+      "loss": 0.2958,
+      "step": 5680
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.717022844509948e-07,
+      "loss": 0.324,
+      "step": 5690
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.7155490051584375e-07,
+      "loss": 0.1269,
+      "step": 5700
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.7140751658069265e-07,
+      "loss": 0.348,
+      "step": 5710
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.712601326455416e-07,
+      "loss": 0.2874,
+      "step": 5720
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.7111274871039055e-07,
+      "loss": 0.2389,
+      "step": 5730
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.7096536477523945e-07,
+      "loss": 0.4025,
+      "step": 5740
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.708179808400884e-07,
+      "loss": 0.2842,
+      "step": 5750
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.7067059690493735e-07,
+      "loss": 0.4168,
+      "step": 5760
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.7052321296978624e-07,
+      "loss": 0.5503,
+      "step": 5770
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.703758290346352e-07,
+      "loss": 0.454,
+      "step": 5780
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.702284450994841e-07,
+      "loss": 0.1809,
+      "step": 5790
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.7008106116433304e-07,
+      "loss": 0.155,
+      "step": 5800
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.69933677229182e-07,
+      "loss": 0.3968,
+      "step": 5810
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.697862932940309e-07,
+      "loss": 0.073,
+      "step": 5820
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.6963890935887984e-07,
+      "loss": 0.3099,
+      "step": 5830
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.694915254237288e-07,
+      "loss": 0.5327,
+      "step": 5840
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.693441414885777e-07,
+      "loss": 0.3137,
+      "step": 5850
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.6919675755342664e-07,
+      "loss": 0.2054,
+      "step": 5860
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.690493736182756e-07,
+      "loss": 0.0266,
+      "step": 5870
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.689019896831245e-07,
+      "loss": 0.3723,
+      "step": 5880
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.6875460574797344e-07,
+      "loss": 0.1546,
+      "step": 5890
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.686072218128224e-07,
+      "loss": 0.4927,
+      "step": 5900
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.684598378776713e-07,
+      "loss": 0.2442,
+      "step": 5910
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.6831245394252024e-07,
+      "loss": 0.4819,
+      "step": 5920
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.6816507000736913e-07,
+      "loss": 0.6341,
+      "step": 5930
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.680176860722181e-07,
+      "loss": 0.2236,
+      "step": 5940
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.6787030213706703e-07,
+      "loss": 0.4526,
+      "step": 5950
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.6772291820191593e-07,
+      "loss": 0.0686,
+      "step": 5960
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.675755342667649e-07,
+      "loss": 0.2784,
+      "step": 5970
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6742815033161383e-07,
+      "loss": 0.1716,
+      "step": 5980
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6728076639646273e-07,
+      "loss": 0.4911,
+      "step": 5990
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.671333824613117e-07,
+      "loss": 0.2251,
+      "step": 6000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6698599852616063e-07,
+      "loss": 0.135,
+      "step": 6010
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6683861459100953e-07,
+      "loss": 0.3255,
+      "step": 6020
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.666912306558585e-07,
+      "loss": 0.2879,
+      "step": 6030
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.665438467207074e-07,
+      "loss": 0.694,
+      "step": 6040
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.663964627855563e-07,
+      "loss": 0.1067,
+      "step": 6050
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.662490788504053e-07,
+      "loss": 0.329,
+      "step": 6060
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.661016949152542e-07,
+      "loss": 0.2664,
+      "step": 6070
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.659543109801031e-07,
+      "loss": 0.6238,
+      "step": 6080
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.658069270449521e-07,
+      "loss": 0.2292,
+      "step": 6090
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6565954310980097e-07,
+      "loss": 0.5738,
+      "step": 6100
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.655121591746499e-07,
+      "loss": 0.3031,
+      "step": 6110
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6536477523949887e-07,
+      "loss": 0.4635,
+      "step": 6120
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6521739130434777e-07,
+      "loss": 0.2825,
+      "step": 6130
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.650700073691967e-07,
+      "loss": 0.4422,
+      "step": 6140
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6492262343404567e-07,
+      "loss": 0.4844,
+      "step": 6150
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6477523949889457e-07,
+      "loss": 0.363,
+      "step": 6160
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.646278555637435e-07,
+      "loss": 0.1012,
+      "step": 6170
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.644804716285924e-07,
+      "loss": 0.4715,
+      "step": 6180
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6433308769344137e-07,
+      "loss": 0.4202,
+      "step": 6190
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.641857037582903e-07,
+      "loss": 0.4372,
+      "step": 6200
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.640383198231392e-07,
+      "loss": 0.4864,
+      "step": 6210
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6389093588798817e-07,
+      "loss": 0.3098,
+      "step": 6220
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.637435519528371e-07,
+      "loss": 0.4132,
+      "step": 6230
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.63596168017686e-07,
+      "loss": 0.5999,
+      "step": 6240
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6344878408253496e-07,
+      "loss": 0.144,
+      "step": 6250
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.633014001473839e-07,
+      "loss": 0.1951,
+      "step": 6260
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.631540162122328e-07,
+      "loss": 0.1672,
+      "step": 6270
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6300663227708176e-07,
+      "loss": 0.4963,
+      "step": 6280
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.628592483419307e-07,
+      "loss": 0.3618,
+      "step": 6290
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.627118644067796e-07,
+      "loss": 0.4101,
+      "step": 6300
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6256448047162856e-07,
+      "loss": 0.1451,
+      "step": 6310
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6241709653647746e-07,
+      "loss": 0.5091,
+      "step": 6320
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.622697126013264e-07,
+      "loss": 0.417,
+      "step": 6330
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.6212232866617536e-07,
+      "loss": 0.474,
+      "step": 6340
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.6197494473102426e-07,
+      "loss": 0.4525,
+      "step": 6350
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.618275607958732e-07,
+      "loss": 0.3916,
+      "step": 6360
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.6168017686072216e-07,
+      "loss": 0.4334,
+      "step": 6370
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.6153279292557105e-07,
+      "loss": 0.2918,
+      "step": 6380
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.6138540899042e-07,
+      "loss": 0.7031,
+      "step": 6390
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.6123802505526895e-07,
+      "loss": 0.3449,
+      "step": 6400
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.6109064112011785e-07,
+      "loss": 0.1435,
+      "step": 6410
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.609432571849668e-07,
+      "loss": 0.2261,
+      "step": 6420
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.607958732498157e-07,
+      "loss": 0.2199,
+      "step": 6430
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.6064848931466465e-07,
+      "loss": 0.3515,
+      "step": 6440
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.605011053795136e-07,
+      "loss": 0.2802,
+      "step": 6450
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.603537214443625e-07,
+      "loss": 0.1916,
+      "step": 6460
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.6020633750921145e-07,
+      "loss": 0.4325,
+      "step": 6470
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.600589535740604e-07,
+      "loss": 0.4674,
+      "step": 6480
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.599115696389093e-07,
+      "loss": 0.0693,
+      "step": 6490
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.5976418570375825e-07,
+      "loss": 0.5154,
+      "step": 6500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.596168017686072e-07,
+      "loss": 0.0863,
+      "step": 6510
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.594694178334561e-07,
+      "loss": 0.4898,
+      "step": 6520
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.5932203389830505e-07,
+      "loss": 0.0885,
+      "step": 6530
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.59174649963154e-07,
+      "loss": 0.5605,
+      "step": 6540
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.590272660280029e-07,
+      "loss": 0.3815,
+      "step": 6550
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.5887988209285184e-07,
+      "loss": 0.0331,
+      "step": 6560
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.5873249815770074e-07,
+      "loss": 0.2007,
+      "step": 6570
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.585851142225497e-07,
+      "loss": 0.452,
+      "step": 6580
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.5843773028739864e-07,
+      "loss": 0.0792,
+      "step": 6590
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.5829034635224754e-07,
+      "loss": 0.3733,
+      "step": 6600
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.581429624170965e-07,
+      "loss": 0.3203,
+      "step": 6610
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.5799557848194544e-07,
+      "loss": 0.0755,
+      "step": 6620
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.5784819454679434e-07,
+      "loss": 0.1678,
+      "step": 6630
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.577008106116433e-07,
+      "loss": 0.0207,
+      "step": 6640
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.5755342667649224e-07,
+      "loss": 0.092,
+      "step": 6650
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.5740604274134114e-07,
+      "loss": 0.6279,
+      "step": 6660
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.572586588061901e-07,
+      "loss": 0.0131,
+      "step": 6670
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.5711127487103904e-07,
+      "loss": 0.0535,
+      "step": 6680
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.5696389093588793e-07,
+      "loss": 0.5263,
+      "step": 6690
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.568165070007369e-07,
+      "loss": 0.7086,
+      "step": 6700
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.566691230655858e-07,
+      "loss": 0.5166,
+      "step": 6710
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.5652173913043473e-07,
+      "loss": 0.2798,
+      "step": 6720
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.563743551952837e-07,
+      "loss": 0.1079,
+      "step": 6730
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.562269712601326e-07,
+      "loss": 0.6789,
+      "step": 6740
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.5607958732498153e-07,
+      "loss": 0.2277,
+      "step": 6750
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.559322033898305e-07,
+      "loss": 0.4831,
+      "step": 6760
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.557848194546794e-07,
+      "loss": 0.8635,
+      "step": 6770
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.5563743551952833e-07,
+      "loss": 0.2406,
+      "step": 6780
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.554900515843773e-07,
+      "loss": 0.0745,
+      "step": 6790
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.553426676492262e-07,
+      "loss": 0.3931,
+      "step": 6800
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.5519528371407513e-07,
+      "loss": 0.2113,
+      "step": 6810
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.55047899778924e-07,
+      "loss": 0.2581,
+      "step": 6820
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.54900515843773e-07,
+      "loss": 0.3694,
+      "step": 6830
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.547531319086219e-07,
+      "loss": 0.278,
+      "step": 6840
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.546057479734708e-07,
+      "loss": 0.1335,
+      "step": 6850
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.544583640383198e-07,
+      "loss": 0.3545,
+      "step": 6860
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.543109801031688e-07,
+      "loss": 0.3536,
+      "step": 6870
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.541635961680177e-07,
+      "loss": 0.4162,
+      "step": 6880
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.540162122328666e-07,
+      "loss": 0.052,
+      "step": 6890
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.538688282977156e-07,
+      "loss": 0.3383,
+      "step": 6900
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.5372144436256447e-07,
+      "loss": 0.7978,
+      "step": 6910
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.535740604274134e-07,
+      "loss": 0.3283,
+      "step": 6920
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.5342667649226237e-07,
+      "loss": 0.4063,
+      "step": 6930
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.5327929255711127e-07,
+      "loss": 0.4052,
+      "step": 6940
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.531319086219602e-07,
+      "loss": 0.4319,
+      "step": 6950
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.529845246868091e-07,
+      "loss": 0.324,
+      "step": 6960
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.5283714075165807e-07,
+      "loss": 0.1557,
+      "step": 6970
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.52689756816507e-07,
+      "loss": 0.3795,
+      "step": 6980
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.525423728813559e-07,
+      "loss": 0.0481,
+      "step": 6990
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.5239498894620487e-07,
+      "loss": 0.3874,
+      "step": 7000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.522476050110538e-07,
+      "loss": 0.1816,
+      "step": 7010
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.521002210759027e-07,
+      "loss": 0.2894,
+      "step": 7020
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.5195283714075166e-07,
+      "loss": 0.174,
+      "step": 7030
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.518054532056006e-07,
+      "loss": 0.1998,
+      "step": 7040
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.516580692704495e-07,
+      "loss": 0.5473,
+      "step": 7050
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.5151068533529846e-07,
+      "loss": 0.2943,
+      "step": 7060
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.513633014001474e-07,
+      "loss": 0.1403,
+      "step": 7070
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.512159174649963e-07,
+      "loss": 0.4682,
+      "step": 7080
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.5106853352984526e-07,
+      "loss": 0.4339,
+      "step": 7090
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.5092114959469416e-07,
+      "loss": 0.1962,
+      "step": 7100
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.507737656595431e-07,
+      "loss": 0.3022,
+      "step": 7110
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.5062638172439206e-07,
+      "loss": 0.1322,
+      "step": 7120
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.5047899778924096e-07,
+      "loss": 0.3952,
+      "step": 7130
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.503316138540899e-07,
+      "loss": 0.1522,
+      "step": 7140
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.5018422991893886e-07,
+      "loss": 0.5476,
+      "step": 7150
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.5003684598378776e-07,
+      "loss": 0.0809,
+      "step": 7160
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.498894620486367e-07,
+      "loss": 0.2128,
+      "step": 7170
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.4974207811348566e-07,
+      "loss": 0.5815,
+      "step": 7180
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.4959469417833455e-07,
+      "loss": 0.3398,
+      "step": 7190
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.494473102431835e-07,
+      "loss": 0.4122,
+      "step": 7200
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.4929992630803245e-07,
+      "loss": 0.2774,
+      "step": 7210
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.4915254237288135e-07,
+      "loss": 0.466,
+      "step": 7220
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.490051584377303e-07,
+      "loss": 0.583,
+      "step": 7230
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.488577745025792e-07,
+      "loss": 0.5025,
+      "step": 7240
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.4871039056742815e-07,
+      "loss": 0.0059,
+      "step": 7250
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.485630066322771e-07,
+      "loss": 0.3219,
+      "step": 7260
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.48415622697126e-07,
+      "loss": 0.0015,
+      "step": 7270
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.4826823876197495e-07,
+      "loss": 0.5898,
+      "step": 7280
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.481208548268239e-07,
+      "loss": 0.0973,
+      "step": 7290
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.479734708916728e-07,
+      "loss": 0.0812,
+      "step": 7300
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.4782608695652175e-07,
+      "loss": 0.8095,
+      "step": 7310
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.476787030213707e-07,
+      "loss": 0.1115,
+      "step": 7320
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.475313190862196e-07,
+      "loss": 0.4173,
+      "step": 7330
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.4738393515106854e-07,
+      "loss": 0.0228,
+      "step": 7340
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.4723655121591744e-07,
+      "loss": 0.466,
+      "step": 7350
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.470891672807664e-07,
+      "loss": 0.2258,
+      "step": 7360
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.4694178334561534e-07,
+      "loss": 0.3372,
+      "step": 7370
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.4679439941046424e-07,
+      "loss": 0.248,
+      "step": 7380
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.466470154753132e-07,
+      "loss": 0.4178,
+      "step": 7390
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.4649963154016214e-07,
+      "loss": 0.3057,
+      "step": 7400
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.4635224760501104e-07,
+      "loss": 0.2706,
+      "step": 7410
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.4620486366986e-07,
+      "loss": 0.9227,
+      "step": 7420
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.4605747973470894e-07,
+      "loss": 0.543,
+      "step": 7430
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.4591009579955784e-07,
+      "loss": 0.0341,
+      "step": 7440
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.457627118644068e-07,
+      "loss": 0.286,
+      "step": 7450
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.4561532792925574e-07,
+      "loss": 0.2712,
+      "step": 7460
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.4546794399410464e-07,
+      "loss": 0.1096,
+      "step": 7470
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.453205600589536e-07,
+      "loss": 0.5792,
+      "step": 7480
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.451731761238025e-07,
+      "loss": 0.7787,
+      "step": 7490
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.4502579218865143e-07,
+      "loss": 0.1855,
+      "step": 7500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.448784082535004e-07,
+      "loss": 0.4726,
+      "step": 7510
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.447310243183493e-07,
+      "loss": 0.1863,
+      "step": 7520
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.4458364038319823e-07,
+      "loss": 0.2212,
+      "step": 7530
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.444362564480472e-07,
+      "loss": 0.1918,
+      "step": 7540
+    },
+    {
+      "epoch": 2.07,
+      "eval_accuracy": 0.8733829870638965,
+      "eval_loss": 0.5665048956871033,
+      "eval_runtime": 617.2351,
+      "eval_samples_per_second": 4.133,
+      "eval_steps_per_second": 1.034,
+      "step": 7542
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.442888725128961e-07,
+      "loss": 0.238,
+      "step": 7550
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.4414148857774503e-07,
+      "loss": 0.1014,
+      "step": 7560
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.43994104642594e-07,
+      "loss": 0.4005,
+      "step": 7570
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.438467207074429e-07,
+      "loss": 0.2532,
+      "step": 7580
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.4369933677229183e-07,
+      "loss": 0.1695,
+      "step": 7590
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.435519528371408e-07,
+      "loss": 0.3114,
+      "step": 7600
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.434045689019897e-07,
+      "loss": 0.4329,
+      "step": 7610
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.432571849668386e-07,
+      "loss": 0.2144,
+      "step": 7620
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.431098010316875e-07,
+      "loss": 0.5157,
+      "step": 7630
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.429624170965365e-07,
+      "loss": 0.3029,
+      "step": 7640
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.428150331613854e-07,
+      "loss": 0.3108,
+      "step": 7650
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.426676492262343e-07,
+      "loss": 0.3325,
+      "step": 7660
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.4252026529108327e-07,
+      "loss": 0.2608,
+      "step": 7670
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.423728813559322e-07,
+      "loss": 0.1544,
+      "step": 7680
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.422254974207811e-07,
+      "loss": 0.327,
+      "step": 7690
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.4207811348563007e-07,
+      "loss": 0.1069,
+      "step": 7700
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.41930729550479e-07,
+      "loss": 0.3136,
+      "step": 7710
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.417833456153279e-07,
+      "loss": 0.6677,
+      "step": 7720
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.4163596168017687e-07,
+      "loss": 0.0038,
+      "step": 7730
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.4148857774502577e-07,
+      "loss": 0.1614,
+      "step": 7740
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.413411938098747e-07,
+      "loss": 0.3297,
+      "step": 7750
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.4119380987472367e-07,
+      "loss": 0.3435,
+      "step": 7760
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.4104642593957256e-07,
+      "loss": 0.2044,
+      "step": 7770
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.408990420044215e-07,
+      "loss": 0.0985,
+      "step": 7780
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.4075165806927047e-07,
+      "loss": 0.4072,
+      "step": 7790
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.4060427413411936e-07,
+      "loss": 0.2261,
+      "step": 7800
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.404568901989683e-07,
+      "loss": 0.1367,
+      "step": 7810
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.4030950626381726e-07,
+      "loss": 0.213,
+      "step": 7820
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.4016212232866616e-07,
+      "loss": 0.1874,
+      "step": 7830
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.400147383935151e-07,
+      "loss": 0.1291,
+      "step": 7840
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.3986735445836406e-07,
+      "loss": 0.0069,
+      "step": 7850
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.3971997052321296e-07,
+      "loss": 0.2835,
+      "step": 7860
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.395725865880619e-07,
+      "loss": 0.3197,
+      "step": 7870
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.394252026529108e-07,
+      "loss": 0.1838,
+      "step": 7880
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.3927781871775976e-07,
+      "loss": 0.0212,
+      "step": 7890
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.391304347826087e-07,
+      "loss": 0.2141,
+      "step": 7900
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.389830508474576e-07,
+      "loss": 0.5995,
+      "step": 7910
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.3883566691230656e-07,
+      "loss": 0.0058,
+      "step": 7920
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.386882829771555e-07,
+      "loss": 0.7008,
+      "step": 7930
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.385408990420044e-07,
+      "loss": 0.3585,
+      "step": 7940
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.3839351510685335e-07,
+      "loss": 0.2724,
+      "step": 7950
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.382461311717023e-07,
+      "loss": 0.3404,
+      "step": 7960
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.380987472365512e-07,
+      "loss": 0.1427,
+      "step": 7970
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.3795136330140015e-07,
+      "loss": 0.5057,
+      "step": 7980
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.378039793662491e-07,
+      "loss": 0.0595,
+      "step": 7990
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.37656595431098e-07,
+      "loss": 0.2981,
+      "step": 8000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.3750921149594695e-07,
+      "loss": 0.3558,
+      "step": 8010
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.3736182756079585e-07,
+      "loss": 0.4554,
+      "step": 8020
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.372144436256448e-07,
+      "loss": 0.231,
+      "step": 8030
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.3706705969049375e-07,
+      "loss": 0.0192,
+      "step": 8040
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.3691967575534265e-07,
+      "loss": 0.2676,
+      "step": 8050
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.367722918201916e-07,
+      "loss": 0.5022,
+      "step": 8060
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.3662490788504055e-07,
+      "loss": 0.0825,
+      "step": 8070
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.3647752394988944e-07,
+      "loss": 0.46,
+      "step": 8080
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.363301400147384e-07,
+      "loss": 0.4752,
+      "step": 8090
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.3618275607958735e-07,
+      "loss": 0.4231,
+      "step": 8100
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.3603537214443624e-07,
+      "loss": 0.9388,
+      "step": 8110
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.358879882092852e-07,
+      "loss": 0.1537,
+      "step": 8120
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.357406042741341e-07,
+      "loss": 0.3789,
+      "step": 8130
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.3559322033898304e-07,
+      "loss": 0.3492,
+      "step": 8140
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.35445836403832e-07,
+      "loss": 0.119,
+      "step": 8150
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.352984524686809e-07,
+      "loss": 0.1383,
+      "step": 8160
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.3515106853352984e-07,
+      "loss": 0.0881,
+      "step": 8170
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.350036845983788e-07,
+      "loss": 0.3361,
+      "step": 8180
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.348563006632277e-07,
+      "loss": 0.1172,
+      "step": 8190
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.3470891672807664e-07,
+      "loss": 0.1064,
+      "step": 8200
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.345615327929256e-07,
+      "loss": 0.13,
+      "step": 8210
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.344141488577745e-07,
+      "loss": 0.3421,
+      "step": 8220
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.3426676492262344e-07,
+      "loss": 0.1586,
+      "step": 8230
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.341193809874724e-07,
+      "loss": 0.1385,
+      "step": 8240
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.339719970523213e-07,
+      "loss": 0.4699,
+      "step": 8250
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.3382461311717023e-07,
+      "loss": 0.03,
+      "step": 8260
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.3367722918201913e-07,
+      "loss": 0.3073,
+      "step": 8270
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.335298452468681e-07,
+      "loss": 0.3121,
+      "step": 8280
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.3338246131171703e-07,
+      "loss": 0.2518,
+      "step": 8290
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.3323507737656593e-07,
+      "loss": 0.4304,
+      "step": 8300
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.330876934414149e-07,
+      "loss": 0.6223,
+      "step": 8310
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.3294030950626383e-07,
+      "loss": 0.1584,
+      "step": 8320
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.3279292557111273e-07,
+      "loss": 0.084,
+      "step": 8330
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.326455416359617e-07,
+      "loss": 0.6036,
+      "step": 8340
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.3249815770081063e-07,
+      "loss": 0.0784,
+      "step": 8350
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.323507737656595e-07,
+      "loss": 0.198,
+      "step": 8360
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.322033898305085e-07,
+      "loss": 0.391,
+      "step": 8370
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.3205600589535743e-07,
+      "loss": 0.1807,
+      "step": 8380
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.319086219602063e-07,
+      "loss": 0.3092,
+      "step": 8390
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.317612380250553e-07,
+      "loss": 0.2035,
+      "step": 8400
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.3161385408990417e-07,
+      "loss": 0.4079,
+      "step": 8410
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.314664701547531e-07,
+      "loss": 0.1219,
+      "step": 8420
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.3131908621960207e-07,
+      "loss": 0.4502,
+      "step": 8430
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.3117170228445097e-07,
+      "loss": 0.0045,
+      "step": 8440
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.310243183492999e-07,
+      "loss": 0.0277,
+      "step": 8450
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.3087693441414887e-07,
+      "loss": 0.4302,
+      "step": 8460
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.3072955047899777e-07,
+      "loss": 0.3171,
+      "step": 8470
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.305821665438467e-07,
+      "loss": 0.3633,
+      "step": 8480
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.3043478260869567e-07,
+      "loss": 0.4951,
+      "step": 8490
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.3028739867354457e-07,
+      "loss": 0.1704,
+      "step": 8500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.301400147383935e-07,
+      "loss": 0.1656,
+      "step": 8510
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.299926308032424e-07,
+      "loss": 0.1575,
+      "step": 8520
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.2984524686809137e-07,
+      "loss": 0.1559,
+      "step": 8530
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.296978629329403e-07,
+      "loss": 0.3115,
+      "step": 8540
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.295504789977892e-07,
+      "loss": 0.2615,
+      "step": 8550
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.2940309506263816e-07,
+      "loss": 0.7415,
+      "step": 8560
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.292557111274871e-07,
+      "loss": 0.2019,
+      "step": 8570
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.29108327192336e-07,
+      "loss": 0.245,
+      "step": 8580
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.2896094325718496e-07,
+      "loss": 1.0986,
+      "step": 8590
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.288135593220339e-07,
+      "loss": 0.0013,
+      "step": 8600
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.286661753868828e-07,
+      "loss": 0.3152,
+      "step": 8610
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.2851879145173176e-07,
+      "loss": 0.1865,
+      "step": 8620
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.283714075165807e-07,
+      "loss": 0.1636,
+      "step": 8630
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.282240235814296e-07,
+      "loss": 0.566,
+      "step": 8640
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.2807663964627856e-07,
+      "loss": 0.2782,
+      "step": 8650
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.2792925571112746e-07,
+      "loss": 0.5805,
+      "step": 8660
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.277818717759764e-07,
+      "loss": 0.3486,
+      "step": 8670
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.2763448784082536e-07,
+      "loss": 0.0836,
+      "step": 8680
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.2748710390567425e-07,
+      "loss": 0.5491,
+      "step": 8690
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.273397199705232e-07,
+      "loss": 0.1537,
+      "step": 8700
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.2719233603537215e-07,
+      "loss": 0.4735,
+      "step": 8710
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.2704495210022105e-07,
+      "loss": 0.458,
+      "step": 8720
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.2689756816507e-07,
+      "loss": 0.2734,
+      "step": 8730
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.2675018422991895e-07,
+      "loss": 0.0856,
+      "step": 8740
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.2660280029476785e-07,
+      "loss": 0.4252,
+      "step": 8750
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.264554163596168e-07,
+      "loss": 0.5923,
+      "step": 8760
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.263080324244657e-07,
+      "loss": 0.3594,
+      "step": 8770
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.2616064848931465e-07,
+      "loss": 0.3816,
+      "step": 8780
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.260132645541636e-07,
+      "loss": 0.1401,
+      "step": 8790
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.258658806190125e-07,
+      "loss": 0.1053,
+      "step": 8800
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.2571849668386145e-07,
+      "loss": 0.618,
+      "step": 8810
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.255711127487104e-07,
+      "loss": 0.1985,
+      "step": 8820
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.254237288135593e-07,
+      "loss": 0.1081,
+      "step": 8830
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.2527634487840825e-07,
+      "loss": 0.5407,
+      "step": 8840
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.251289609432572e-07,
+      "loss": 0.0384,
+      "step": 8850
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.249815770081061e-07,
+      "loss": 0.0384,
+      "step": 8860
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.2483419307295504e-07,
+      "loss": 0.143,
+      "step": 8870
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.24686809137804e-07,
+      "loss": 0.1811,
+      "step": 8880
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.245394252026529e-07,
+      "loss": 0.4314,
+      "step": 8890
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.2439204126750184e-07,
+      "loss": 0.2934,
+      "step": 8900
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.2424465733235074e-07,
+      "loss": 0.3258,
+      "step": 8910
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.240972733971997e-07,
+      "loss": 0.4801,
+      "step": 8920
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.2394988946204864e-07,
+      "loss": 0.2329,
+      "step": 8930
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.2380250552689754e-07,
+      "loss": 0.0612,
+      "step": 8940
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.236551215917465e-07,
+      "loss": 0.2026,
+      "step": 8950
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.2350773765659544e-07,
+      "loss": 0.2761,
+      "step": 8960
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.2336035372144434e-07,
+      "loss": 0.2373,
+      "step": 8970
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.232129697862933e-07,
+      "loss": 0.0295,
+      "step": 8980
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.2306558585114224e-07,
+      "loss": 0.1401,
+      "step": 8990
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.2291820191599113e-07,
+      "loss": 0.0365,
+      "step": 9000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.227708179808401e-07,
+      "loss": 0.1278,
+      "step": 9010
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.2262343404568903e-07,
+      "loss": 0.3538,
+      "step": 9020
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.2247605011053793e-07,
+      "loss": 0.0955,
+      "step": 9030
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.223286661753869e-07,
+      "loss": 0.214,
+      "step": 9040
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.221812822402358e-07,
+      "loss": 0.2707,
+      "step": 9050
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.2203389830508473e-07,
+      "loss": 0.2139,
+      "step": 9060
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.218865143699337e-07,
+      "loss": 0.2709,
+      "step": 9070
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.217391304347826e-07,
+      "loss": 0.4966,
+      "step": 9080
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.2159174649963153e-07,
+      "loss": 0.5115,
+      "step": 9090
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.214443625644805e-07,
+      "loss": 0.0156,
+      "step": 9100
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.212969786293294e-07,
+      "loss": 0.8871,
+      "step": 9110
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.2114959469417833e-07,
+      "loss": 0.4056,
+      "step": 9120
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.210022107590273e-07,
+      "loss": 0.1822,
+      "step": 9130
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.208548268238762e-07,
+      "loss": 0.3387,
+      "step": 9140
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.207074428887251e-07,
+      "loss": 0.0086,
+      "step": 9150
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.20560058953574e-07,
+      "loss": 0.0851,
+      "step": 9160
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.2041267501842297e-07,
+      "loss": 0.3681,
+      "step": 9170
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.202652910832719e-07,
+      "loss": 0.002,
+      "step": 9180
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.201179071481208e-07,
+      "loss": 0.1719,
+      "step": 9190
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.1997052321296977e-07,
+      "loss": 0.1541,
+      "step": 9200
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.198231392778187e-07,
+      "loss": 0.0943,
+      "step": 9210
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.196757553426676e-07,
+      "loss": 0.2001,
+      "step": 9220
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.1952837140751657e-07,
+      "loss": 0.1583,
+      "step": 9230
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.193809874723655e-07,
+      "loss": 0.2131,
+      "step": 9240
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.192336035372144e-07,
+      "loss": 0.439,
+      "step": 9250
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.1908621960206337e-07,
+      "loss": 0.107,
+      "step": 9260
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.189388356669123e-07,
+      "loss": 0.1905,
+      "step": 9270
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.187914517317612e-07,
+      "loss": 0.0108,
+      "step": 9280
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.1864406779661017e-07,
+      "loss": 0.2254,
+      "step": 9290
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.1849668386145906e-07,
+      "loss": 0.4028,
+      "step": 9300
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.18349299926308e-07,
+      "loss": 0.1872,
+      "step": 9310
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.1820191599115696e-07,
+      "loss": 0.0012,
+      "step": 9320
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.1805453205600586e-07,
+      "loss": 0.2445,
+      "step": 9330
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.179071481208548e-07,
+      "loss": 0.4392,
+      "step": 9340
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.1775976418570376e-07,
+      "loss": 0.111,
+      "step": 9350
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.1761238025055266e-07,
+      "loss": 0.0337,
+      "step": 9360
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.174649963154016e-07,
+      "loss": 0.226,
+      "step": 9370
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.1731761238025056e-07,
+      "loss": 0.348,
+      "step": 9380
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.1717022844509946e-07,
+      "loss": 0.0008,
+      "step": 9390
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.170228445099484e-07,
+      "loss": 0.1891,
+      "step": 9400
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.1687546057479736e-07,
+      "loss": 0.0008,
+      "step": 9410
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.1672807663964626e-07,
+      "loss": 0.2506,
+      "step": 9420
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.165806927044952e-07,
+      "loss": 0.3626,
+      "step": 9430
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.164333087693441e-07,
+      "loss": 0.4256,
+      "step": 9440
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.1628592483419305e-07,
+      "loss": 0.2525,
+      "step": 9450
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.16138540899042e-07,
+      "loss": 0.2063,
+      "step": 9460
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.159911569638909e-07,
+      "loss": 0.5233,
+      "step": 9470
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.1584377302873985e-07,
+      "loss": 0.226,
+      "step": 9480
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.156963890935888e-07,
+      "loss": 0.0015,
+      "step": 9490
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.155490051584377e-07,
+      "loss": 0.33,
+      "step": 9500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.1540162122328665e-07,
+      "loss": 0.1275,
+      "step": 9510
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.152542372881356e-07,
+      "loss": 0.3788,
+      "step": 9520
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.151068533529845e-07,
+      "loss": 0.3547,
+      "step": 9530
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.1495946941783345e-07,
+      "loss": 0.6407,
+      "step": 9540
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.1481208548268235e-07,
+      "loss": 0.0059,
+      "step": 9550
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.146647015475313e-07,
+      "loss": 0.2599,
+      "step": 9560
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.1451731761238025e-07,
+      "loss": 0.0339,
+      "step": 9570
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.1436993367722914e-07,
+      "loss": 0.3814,
+      "step": 9580
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.142225497420781e-07,
+      "loss": 0.2984,
+      "step": 9590
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.1407516580692705e-07,
+      "loss": 0.0055,
+      "step": 9600
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.1392778187177594e-07,
+      "loss": 0.4123,
+      "step": 9610
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.137803979366249e-07,
+      "loss": 0.2254,
+      "step": 9620
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.1363301400147384e-07,
+      "loss": 0.5009,
+      "step": 9630
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.1348563006632274e-07,
+      "loss": 0.2131,
+      "step": 9640
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.133382461311717e-07,
+      "loss": 0.2322,
+      "step": 9650
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.1319086219602064e-07,
+      "loss": 0.302,
+      "step": 9660
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.1304347826086954e-07,
+      "loss": 0.0459,
+      "step": 9670
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.128960943257185e-07,
+      "loss": 0.2679,
+      "step": 9680
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.127487103905674e-07,
+      "loss": 0.4142,
+      "step": 9690
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.1260132645541634e-07,
+      "loss": 0.224,
+      "step": 9700
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.124539425202653e-07,
+      "loss": 0.0249,
+      "step": 9710
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.123065585851142e-07,
+      "loss": 0.5292,
+      "step": 9720
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.1215917464996314e-07,
+      "loss": 0.3092,
+      "step": 9730
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.120117907148121e-07,
+      "loss": 0.3395,
+      "step": 9740
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.11864406779661e-07,
+      "loss": 0.5832,
+      "step": 9750
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.1171702284450993e-07,
+      "loss": 0.0278,
+      "step": 9760
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.115696389093589e-07,
+      "loss": 0.1905,
+      "step": 9770
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.114222549742078e-07,
+      "loss": 0.2142,
+      "step": 9780
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.1127487103905673e-07,
+      "loss": 0.3206,
+      "step": 9790
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.111274871039057e-07,
+      "loss": 0.0022,
+      "step": 9800
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.109801031687546e-07,
+      "loss": 0.2186,
+      "step": 9810
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.1083271923360353e-07,
+      "loss": 0.3053,
+      "step": 9820
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.1068533529845243e-07,
+      "loss": 0.3002,
+      "step": 9830
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.105379513633014e-07,
+      "loss": 0.0223,
+      "step": 9840
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.1039056742815033e-07,
+      "loss": 0.5033,
+      "step": 9850
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.1024318349299923e-07,
+      "loss": 0.5327,
+      "step": 9860
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.100957995578482e-07,
+      "loss": 0.0239,
+      "step": 9870
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.0994841562269713e-07,
+      "loss": 0.389,
+      "step": 9880
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.09801031687546e-07,
+      "loss": 0.1463,
+      "step": 9890
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.09653647752395e-07,
+      "loss": 0.2887,
+      "step": 9900
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.095062638172439e-07,
+      "loss": 0.4023,
+      "step": 9910
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.093588798820928e-07,
+      "loss": 0.311,
+      "step": 9920
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.0921149594694177e-07,
+      "loss": 0.4316,
+      "step": 9930
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.0906411201179067e-07,
+      "loss": 0.3429,
+      "step": 9940
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.089167280766396e-07,
+      "loss": 0.1051,
+      "step": 9950
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.0876934414148857e-07,
+      "loss": 0.2094,
+      "step": 9960
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.0862196020633747e-07,
+      "loss": 0.0672,
+      "step": 9970
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.084745762711864e-07,
+      "loss": 0.3788,
+      "step": 9980
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.0832719233603537e-07,
+      "loss": 0.2211,
+      "step": 9990
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.0817980840088427e-07,
+      "loss": 0.1666,
+      "step": 10000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.080324244657332e-07,
+      "loss": 0.0509,
+      "step": 10010
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.0788504053058217e-07,
+      "loss": 0.0014,
+      "step": 10020
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.0773765659543107e-07,
+      "loss": 0.1289,
+      "step": 10030
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.0759027266028e-07,
+      "loss": 0.0188,
+      "step": 10040
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.0744288872512897e-07,
+      "loss": 0.2002,
+      "step": 10050
+    },
+    {
+      "epoch": 3.07,
+      "eval_accuracy": 0.8808310466483732,
+      "eval_loss": 0.6057535409927368,
+      "eval_runtime": 604.2019,
+      "eval_samples_per_second": 4.222,
+      "eval_steps_per_second": 1.056,
+      "step": 10056
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.0729550478997786e-07,
+      "loss": 0.0954,
+      "step": 10060
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.071481208548268e-07,
+      "loss": 0.081,
+      "step": 10070
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.070007369196757e-07,
+      "loss": 0.0707,
+      "step": 10080
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.0685335298452466e-07,
+      "loss": 0.1897,
+      "step": 10090
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.067059690493736e-07,
+      "loss": 0.2135,
+      "step": 10100
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.065585851142225e-07,
+      "loss": 0.3736,
+      "step": 10110
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.0641120117907146e-07,
+      "loss": 0.1474,
+      "step": 10120
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.062638172439204e-07,
+      "loss": 0.0916,
+      "step": 10130
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.061164333087693e-07,
+      "loss": 0.3906,
+      "step": 10140
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.0596904937361826e-07,
+      "loss": 0.4113,
+      "step": 10150
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.058216654384672e-07,
+      "loss": 0.174,
+      "step": 10160
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.056742815033161e-07,
+      "loss": 0.1678,
+      "step": 10170
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.0552689756816506e-07,
+      "loss": 0.0109,
+      "step": 10180
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.05379513633014e-07,
+      "loss": 0.4722,
+      "step": 10190
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.052321296978629e-07,
+      "loss": 0.5107,
+      "step": 10200
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.0508474576271186e-07,
+      "loss": 0.3098,
+      "step": 10210
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.0493736182756075e-07,
+      "loss": 0.224,
+      "step": 10220
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.047899778924097e-07,
+      "loss": 0.1272,
+      "step": 10230
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.0464259395725865e-07,
+      "loss": 0.2552,
+      "step": 10240
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.0449521002210755e-07,
+      "loss": 0.223,
+      "step": 10250
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.043478260869565e-07,
+      "loss": 0.4706,
+      "step": 10260
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.0420044215180545e-07,
+      "loss": 0.4951,
+      "step": 10270
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.0405305821665435e-07,
+      "loss": 0.2253,
+      "step": 10280
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.039056742815033e-07,
+      "loss": 0.1102,
+      "step": 10290
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.0375829034635225e-07,
+      "loss": 0.1085,
+      "step": 10300
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.0361090641120115e-07,
+      "loss": 0.0727,
+      "step": 10310
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.034635224760501e-07,
+      "loss": 0.3687,
+      "step": 10320
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.03316138540899e-07,
+      "loss": 0.1489,
+      "step": 10330
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.0316875460574795e-07,
+      "loss": 0.0175,
+      "step": 10340
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.030213706705969e-07,
+      "loss": 0.1789,
+      "step": 10350
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.028739867354458e-07,
+      "loss": 0.2648,
+      "step": 10360
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.0272660280029474e-07,
+      "loss": 0.3106,
+      "step": 10370
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.025792188651437e-07,
+      "loss": 0.019,
+      "step": 10380
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.024318349299926e-07,
+      "loss": 0.4892,
+      "step": 10390
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.0228445099484154e-07,
+      "loss": 0.1833,
+      "step": 10400
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.021370670596905e-07,
+      "loss": 0.3526,
+      "step": 10410
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.019896831245394e-07,
+      "loss": 0.0007,
+      "step": 10420
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.0184229918938834e-07,
+      "loss": 0.0021,
+      "step": 10430
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.016949152542373e-07,
+      "loss": 0.0398,
+      "step": 10440
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.015475313190862e-07,
+      "loss": 0.3147,
+      "step": 10450
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.0140014738393514e-07,
+      "loss": 0.5387,
+      "step": 10460
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.0125276344878404e-07,
+      "loss": 0.1767,
+      "step": 10470
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.01105379513633e-07,
+      "loss": 0.1614,
+      "step": 10480
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.0095799557848194e-07,
+      "loss": 0.2212,
+      "step": 10490
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.0081061164333083e-07,
+      "loss": 0.2037,
+      "step": 10500
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.006632277081798e-07,
+      "loss": 0.51,
+      "step": 10510
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.0051584377302873e-07,
+      "loss": 0.3607,
+      "step": 10520
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.0036845983787763e-07,
+      "loss": 0.0008,
+      "step": 10530
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.002210759027266e-07,
+      "loss": 0.3942,
+      "step": 10540
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.0007369196757553e-07,
+      "loss": 0.2379,
+      "step": 10550
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.9992630803242443e-07,
+      "loss": 0.4363,
+      "step": 10560
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.997789240972734e-07,
+      "loss": 0.1681,
+      "step": 10570
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.996315401621223e-07,
+      "loss": 0.0013,
+      "step": 10580
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.9948415622697123e-07,
+      "loss": 0.354,
+      "step": 10590
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.993367722918202e-07,
+      "loss": 0.0833,
+      "step": 10600
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.991893883566691e-07,
+      "loss": 0.349,
+      "step": 10610
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.9904200442151803e-07,
+      "loss": 0.0426,
+      "step": 10620
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.98894620486367e-07,
+      "loss": 0.3468,
+      "step": 10630
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.987472365512159e-07,
+      "loss": 0.0551,
+      "step": 10640
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.985998526160648e-07,
+      "loss": 0.1622,
+      "step": 10650
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.984524686809138e-07,
+      "loss": 0.1751,
+      "step": 10660
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.9830508474576267e-07,
+      "loss": 0.2835,
+      "step": 10670
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.981577008106116e-07,
+      "loss": 0.6271,
+      "step": 10680
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.980103168754606e-07,
+      "loss": 0.0078,
+      "step": 10690
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.9786293294030947e-07,
+      "loss": 0.1163,
+      "step": 10700
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.977155490051584e-07,
+      "loss": 0.0034,
+      "step": 10710
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.975681650700073e-07,
+      "loss": 0.2311,
+      "step": 10720
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.9742078113485627e-07,
+      "loss": 0.0008,
+      "step": 10730
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.972733971997052e-07,
+      "loss": 0.0021,
+      "step": 10740
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.971260132645541e-07,
+      "loss": 0.1703,
+      "step": 10750
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.9697862932940307e-07,
+      "loss": 0.242,
+      "step": 10760
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.96831245394252e-07,
+      "loss": 0.0039,
+      "step": 10770
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.966838614591009e-07,
+      "loss": 0.0956,
+      "step": 10780
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.9653647752394987e-07,
+      "loss": 0.009,
+      "step": 10790
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.963890935887988e-07,
+      "loss": 0.1423,
+      "step": 10800
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.962417096536477e-07,
+      "loss": 0.0006,
+      "step": 10810
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.9609432571849666e-07,
+      "loss": 0.0056,
+      "step": 10820
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.959469417833456e-07,
+      "loss": 0.0015,
+      "step": 10830
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.957995578481945e-07,
+      "loss": 0.1351,
+      "step": 10840
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.9565217391304346e-07,
+      "loss": 0.034,
+      "step": 10850
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.9550478997789236e-07,
+      "loss": 0.0008,
+      "step": 10860
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.953574060427413e-07,
+      "loss": 0.353,
+      "step": 10870
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.9521002210759026e-07,
+      "loss": 0.1533,
+      "step": 10880
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.9506263817243916e-07,
+      "loss": 0.1471,
+      "step": 10890
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.949152542372881e-07,
+      "loss": 0.1746,
+      "step": 10900
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.9476787030213706e-07,
+      "loss": 0.3841,
+      "step": 10910
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.9462048636698596e-07,
+      "loss": 0.0007,
+      "step": 10920
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.944731024318349e-07,
+      "loss": 0.0079,
+      "step": 10930
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.9432571849668386e-07,
+      "loss": 0.0033,
+      "step": 10940
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.9417833456153275e-07,
+      "loss": 0.0072,
+      "step": 10950
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.940309506263817e-07,
+      "loss": 0.2714,
+      "step": 10960
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.938835666912306e-07,
+      "loss": 0.5275,
+      "step": 10970
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.9373618275607955e-07,
+      "loss": 0.0005,
+      "step": 10980
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.935887988209285e-07,
+      "loss": 0.436,
+      "step": 10990
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.934414148857774e-07,
+      "loss": 0.5143,
+      "step": 11000
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.9329403095062635e-07,
+      "loss": 0.7777,
+      "step": 11010
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.931466470154753e-07,
+      "loss": 0.2243,
+      "step": 11020
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.929992630803242e-07,
+      "loss": 0.1111,
+      "step": 11030
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.9285187914517315e-07,
+      "loss": 0.31,
+      "step": 11040
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.927044952100221e-07,
+      "loss": 0.0049,
+      "step": 11050
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.92557111274871e-07,
+      "loss": 0.4244,
+      "step": 11060
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.9240972733971995e-07,
+      "loss": 0.0053,
+      "step": 11070
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.922623434045689e-07,
+      "loss": 0.0085,
+      "step": 11080
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.921149594694178e-07,
+      "loss": 0.0598,
+      "step": 11090
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.9196757553426675e-07,
+      "loss": 0.0011,
+      "step": 11100
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.9182019159911564e-07,
+      "loss": 0.5233,
+      "step": 11110
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.916728076639646e-07,
+      "loss": 0.0073,
+      "step": 11120
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.9152542372881354e-07,
+      "loss": 0.493,
+      "step": 11130
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.9137803979366244e-07,
+      "loss": 0.2545,
+      "step": 11140
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.912306558585114e-07,
+      "loss": 0.1798,
+      "step": 11150
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.9108327192336034e-07,
+      "loss": 0.1716,
+      "step": 11160
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.9093588798820924e-07,
+      "loss": 0.0007,
+      "step": 11170
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.907885040530582e-07,
+      "loss": 0.1961,
+      "step": 11180
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.9064112011790714e-07,
+      "loss": 0.2454,
+      "step": 11190
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.9049373618275604e-07,
+      "loss": 0.0323,
+      "step": 11200
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.90346352247605e-07,
+      "loss": 0.302,
+      "step": 11210
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.9019896831245394e-07,
+      "loss": 0.5046,
+      "step": 11220
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.9005158437730284e-07,
+      "loss": 0.0168,
+      "step": 11230
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.899042004421518e-07,
+      "loss": 0.5013,
+      "step": 11240
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.897568165070007e-07,
+      "loss": 0.0024,
+      "step": 11250
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.8960943257184963e-07,
+      "loss": 0.1895,
+      "step": 11260
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.894620486366986e-07,
+      "loss": 0.3341,
+      "step": 11270
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.893146647015475e-07,
+      "loss": 0.2104,
+      "step": 11280
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.8916728076639643e-07,
+      "loss": 0.0169,
+      "step": 11290
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.890198968312454e-07,
+      "loss": 0.1601,
+      "step": 11300
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.888725128960943e-07,
+      "loss": 0.395,
+      "step": 11310
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.8872512896094323e-07,
+      "loss": 0.0184,
+      "step": 11320
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.885777450257922e-07,
+      "loss": 0.0191,
+      "step": 11330
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.884303610906411e-07,
+      "loss": 0.003,
+      "step": 11340
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.8828297715549003e-07,
+      "loss": 0.2458,
+      "step": 11350
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.8813559322033893e-07,
+      "loss": 0.1777,
+      "step": 11360
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.879882092851879e-07,
+      "loss": 0.0076,
+      "step": 11370
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.8784082535003683e-07,
+      "loss": 0.2758,
+      "step": 11380
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.876934414148857e-07,
+      "loss": 0.0049,
+      "step": 11390
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.875460574797347e-07,
+      "loss": 0.1476,
+      "step": 11400
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.873986735445836e-07,
+      "loss": 0.1663,
+      "step": 11410
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.872512896094325e-07,
+      "loss": 0.1476,
+      "step": 11420
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.871039056742815e-07,
+      "loss": 0.1223,
+      "step": 11430
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.869565217391304e-07,
+      "loss": 0.204,
+      "step": 11440
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.868091378039793e-07,
+      "loss": 0.363,
+      "step": 11450
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.8666175386882827e-07,
+      "loss": 0.532,
+      "step": 11460
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.865143699336772e-07,
+      "loss": 0.1038,
+      "step": 11470
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.863669859985261e-07,
+      "loss": 0.4655,
+      "step": 11480
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.8621960206337507e-07,
+      "loss": 0.33,
+      "step": 11490
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.8607221812822397e-07,
+      "loss": 0.0887,
+      "step": 11500
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.859248341930729e-07,
+      "loss": 0.1865,
+      "step": 11510
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.8577745025792187e-07,
+      "loss": 0.0007,
+      "step": 11520
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.8563006632277077e-07,
+      "loss": 0.095,
+      "step": 11530
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.854826823876197e-07,
+      "loss": 0.4602,
+      "step": 11540
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.8533529845246867e-07,
+      "loss": 0.3431,
+      "step": 11550
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.8518791451731756e-07,
+      "loss": 0.1619,
+      "step": 11560
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.850405305821665e-07,
+      "loss": 0.3282,
+      "step": 11570
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.8489314664701546e-07,
+      "loss": 0.3506,
+      "step": 11580
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.8474576271186436e-07,
+      "loss": 0.065,
+      "step": 11590
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.845983787767133e-07,
+      "loss": 0.2651,
+      "step": 11600
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.8445099484156226e-07,
+      "loss": 0.3091,
+      "step": 11610
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.8430361090641116e-07,
+      "loss": 0.0006,
+      "step": 11620
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.841562269712601e-07,
+      "loss": 0.1738,
+      "step": 11630
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.84008843036109e-07,
+      "loss": 0.1468,
+      "step": 11640
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.8386145910095796e-07,
+      "loss": 0.1982,
+      "step": 11650
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.837140751658069e-07,
+      "loss": 0.0065,
+      "step": 11660
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.835666912306558e-07,
+      "loss": 0.1684,
+      "step": 11670
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.8341930729550476e-07,
+      "loss": 0.0473,
+      "step": 11680
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.832719233603537e-07,
+      "loss": 0.2952,
+      "step": 11690
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.831245394252026e-07,
+      "loss": 0.1979,
+      "step": 11700
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.8297715549005156e-07,
+      "loss": 0.2235,
+      "step": 11710
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.828297715549005e-07,
+      "loss": 0.5803,
+      "step": 11720
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.826823876197494e-07,
+      "loss": 0.1256,
+      "step": 11730
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.8253500368459835e-07,
+      "loss": 0.2843,
+      "step": 11740
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.8238761974944725e-07,
+      "loss": 0.0036,
+      "step": 11750
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.822402358142962e-07,
+      "loss": 0.0057,
+      "step": 11760
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.8209285187914515e-07,
+      "loss": 0.1556,
+      "step": 11770
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.8194546794399405e-07,
+      "loss": 0.4593,
+      "step": 11780
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.81798084008843e-07,
+      "loss": 0.3135,
+      "step": 11790
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.8165070007369195e-07,
+      "loss": 0.2902,
+      "step": 11800
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.8150331613854085e-07,
+      "loss": 0.3186,
+      "step": 11810
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.813559322033898e-07,
+      "loss": 0.0066,
+      "step": 11820
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.8120854826823875e-07,
+      "loss": 0.0005,
+      "step": 11830
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.8106116433308765e-07,
+      "loss": 0.1147,
+      "step": 11840
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.809137803979366e-07,
+      "loss": 0.6574,
+      "step": 11850
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.8076639646278555e-07,
+      "loss": 0.0217,
+      "step": 11860
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.8061901252763444e-07,
+      "loss": 0.0852,
+      "step": 11870
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.804716285924834e-07,
+      "loss": 0.2569,
+      "step": 11880
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.803242446573323e-07,
+      "loss": 0.3721,
+      "step": 11890
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.8017686072218124e-07,
+      "loss": 0.1705,
+      "step": 11900
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.800294767870302e-07,
+      "loss": 0.513,
+      "step": 11910
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.798820928518791e-07,
+      "loss": 0.0898,
+      "step": 11920
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.7973470891672804e-07,
+      "loss": 0.3014,
+      "step": 11930
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.79587324981577e-07,
+      "loss": 0.0436,
+      "step": 11940
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.794399410464259e-07,
+      "loss": 0.001,
+      "step": 11950
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.7929255711127484e-07,
+      "loss": 0.0403,
+      "step": 11960
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.791451731761238e-07,
+      "loss": 0.3402,
+      "step": 11970
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.789977892409727e-07,
+      "loss": 0.0525,
+      "step": 11980
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.7885040530582164e-07,
+      "loss": 0.2848,
+      "step": 11990
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.787030213706706e-07,
+      "loss": 0.0005,
+      "step": 12000
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.785556374355195e-07,
+      "loss": 0.1316,
+      "step": 12010
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.7840825350036844e-07,
+      "loss": 0.1785,
+      "step": 12020
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.7826086956521733e-07,
+      "loss": 0.1965,
+      "step": 12030
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.781134856300663e-07,
+      "loss": 0.1402,
+      "step": 12040
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.7796610169491523e-07,
+      "loss": 0.1439,
+      "step": 12050
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.7781871775976413e-07,
+      "loss": 0.1475,
+      "step": 12060
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.776713338246131e-07,
+      "loss": 0.2888,
+      "step": 12070
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.7752394988946203e-07,
+      "loss": 0.1648,
+      "step": 12080
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.7737656595431093e-07,
+      "loss": 0.1902,
+      "step": 12090
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.772291820191599e-07,
+      "loss": 0.3355,
+      "step": 12100
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.7708179808400883e-07,
+      "loss": 0.2693,
+      "step": 12110
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.7693441414885773e-07,
+      "loss": 0.2979,
+      "step": 12120
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.767870302137067e-07,
+      "loss": 0.1887,
+      "step": 12130
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.766396462785556e-07,
+      "loss": 0.2398,
+      "step": 12140
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.764922623434045e-07,
+      "loss": 0.1594,
+      "step": 12150
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.763448784082535e-07,
+      "loss": 0.0012,
+      "step": 12160
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.761974944731024e-07,
+      "loss": 0.097,
+      "step": 12170
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.760501105379513e-07,
+      "loss": 0.0011,
+      "step": 12180
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.759027266028003e-07,
+      "loss": 0.4986,
+      "step": 12190
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.7575534266764917e-07,
+      "loss": 0.411,
+      "step": 12200
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.756079587324981e-07,
+      "loss": 0.0417,
+      "step": 12210
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.7546057479734707e-07,
+      "loss": 0.6189,
+      "step": 12220
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.7531319086219597e-07,
+      "loss": 0.002,
+      "step": 12230
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.751658069270449e-07,
+      "loss": 0.2726,
+      "step": 12240
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.7501842299189387e-07,
+      "loss": 0.0824,
+      "step": 12250
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.7487103905674277e-07,
+      "loss": 0.0356,
+      "step": 12260
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.747236551215917e-07,
+      "loss": 0.426,
+      "step": 12270
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.745762711864406e-07,
+      "loss": 0.0644,
+      "step": 12280
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.7442888725128957e-07,
+      "loss": 0.3313,
+      "step": 12290
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.742815033161385e-07,
+      "loss": 0.1646,
+      "step": 12300
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.741341193809874e-07,
+      "loss": 0.3781,
+      "step": 12310
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.7398673544583636e-07,
+      "loss": 0.0023,
+      "step": 12320
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.738393515106853e-07,
+      "loss": 0.4428,
+      "step": 12330
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.736919675755342e-07,
+      "loss": 0.0009,
+      "step": 12340
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.7354458364038316e-07,
+      "loss": 0.1356,
+      "step": 12350
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.733971997052321e-07,
+      "loss": 0.2031,
+      "step": 12360
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.73249815770081e-07,
+      "loss": 0.1471,
+      "step": 12370
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.7310243183492996e-07,
+      "loss": 0.1632,
+      "step": 12380
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.7295504789977886e-07,
+      "loss": 0.0061,
+      "step": 12390
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.728076639646278e-07,
+      "loss": 0.3843,
+      "step": 12400
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.7266028002947676e-07,
+      "loss": 0.3313,
+      "step": 12410
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.7251289609432566e-07,
+      "loss": 0.0173,
+      "step": 12420
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.723655121591746e-07,
+      "loss": 0.1536,
+      "step": 12430
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.7221812822402356e-07,
+      "loss": 0.1945,
+      "step": 12440
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.7207074428887246e-07,
+      "loss": 0.1325,
+      "step": 12450
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.719233603537214e-07,
+      "loss": 0.1411,
+      "step": 12460
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.7177597641857036e-07,
+      "loss": 0.3863,
+      "step": 12470
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.7162859248341925e-07,
+      "loss": 0.0915,
+      "step": 12480
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.714812085482682e-07,
+      "loss": 0.2071,
+      "step": 12490
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.7133382461311715e-07,
+      "loss": 0.5152,
+      "step": 12500
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.7118644067796605e-07,
+      "loss": 0.1386,
+      "step": 12510
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.71039056742815e-07,
+      "loss": 0.5015,
+      "step": 12520
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.708916728076639e-07,
+      "loss": 0.2727,
+      "step": 12530
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.7074428887251285e-07,
+      "loss": 0.0428,
+      "step": 12540
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.705969049373618e-07,
+      "loss": 0.0021,
+      "step": 12550
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.704495210022107e-07,
+      "loss": 0.3366,
+      "step": 12560
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.7030213706705965e-07,
+      "loss": 0.1921,
+      "step": 12570
+    },
+    {
+      "epoch": 4.07,
+      "eval_accuracy": 0.8867110936887496,
+      "eval_loss": 0.6136470437049866,
+      "eval_runtime": 593.6982,
+      "eval_samples_per_second": 4.297,
+      "eval_steps_per_second": 1.075,
+      "step": 12570
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.701547531319086e-07,
+      "loss": 0.2405,
+      "step": 12580
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.700073691967575e-07,
+      "loss": 0.0003,
+      "step": 12590
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.6985998526160645e-07,
+      "loss": 0.3678,
+      "step": 12600
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.697126013264554e-07,
+      "loss": 0.0699,
+      "step": 12610
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.695652173913043e-07,
+      "loss": 0.1983,
+      "step": 12620
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.6941783345615324e-07,
+      "loss": 0.4901,
+      "step": 12630
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.692704495210022e-07,
+      "loss": 0.1043,
+      "step": 12640
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.691230655858511e-07,
+      "loss": 0.0005,
+      "step": 12650
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.6897568165070004e-07,
+      "loss": 0.1684,
+      "step": 12660
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.6882829771554894e-07,
+      "loss": 0.0839,
+      "step": 12670
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.686809137803979e-07,
+      "loss": 0.0302,
+      "step": 12680
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.6853352984524684e-07,
+      "loss": 0.006,
+      "step": 12690
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.6838614591009574e-07,
+      "loss": 0.0094,
+      "step": 12700
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.682387619749447e-07,
+      "loss": 0.1905,
+      "step": 12710
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.6809137803979364e-07,
+      "loss": 0.1789,
+      "step": 12720
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.6794399410464254e-07,
+      "loss": 0.3499,
+      "step": 12730
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.677966101694915e-07,
+      "loss": 0.4747,
+      "step": 12740
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.6764922623434044e-07,
+      "loss": 0.1692,
+      "step": 12750
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.6750184229918934e-07,
+      "loss": 0.0003,
+      "step": 12760
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.673544583640383e-07,
+      "loss": 0.258,
+      "step": 12770
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.672070744288872e-07,
+      "loss": 0.0018,
+      "step": 12780
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.6705969049373613e-07,
+      "loss": 0.0217,
+      "step": 12790
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.669123065585851e-07,
+      "loss": 0.2134,
+      "step": 12800
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.66764922623434e-07,
+      "loss": 0.2296,
+      "step": 12810
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.6661753868828293e-07,
+      "loss": 0.2651,
+      "step": 12820
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.664701547531319e-07,
+      "loss": 0.3355,
+      "step": 12830
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.663227708179808e-07,
+      "loss": 0.2962,
+      "step": 12840
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.6617538688282973e-07,
+      "loss": 0.3061,
+      "step": 12850
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.660280029476787e-07,
+      "loss": 0.0126,
+      "step": 12860
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.658806190125276e-07,
+      "loss": 0.0002,
+      "step": 12870
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.6573323507737653e-07,
+      "loss": 0.1857,
+      "step": 12880
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.655858511422255e-07,
+      "loss": 0.1672,
+      "step": 12890
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.654384672070744e-07,
+      "loss": 0.2705,
+      "step": 12900
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.652910832719233e-07,
+      "loss": 0.347,
+      "step": 12910
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.651436993367722e-07,
+      "loss": 0.3589,
+      "step": 12920
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.649963154016212e-07,
+      "loss": 0.3215,
+      "step": 12930
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.648489314664701e-07,
+      "loss": 0.1603,
+      "step": 12940
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.64701547531319e-07,
+      "loss": 0.0013,
+      "step": 12950
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.6455416359616797e-07,
+      "loss": 0.0799,
+      "step": 12960
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.644067796610169e-07,
+      "loss": 0.0872,
+      "step": 12970
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.642593957258658e-07,
+      "loss": 0.0005,
+      "step": 12980
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.6411201179071477e-07,
+      "loss": 0.0032,
+      "step": 12990
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.639646278555637e-07,
+      "loss": 0.1438,
+      "step": 13000
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.638172439204126e-07,
+      "loss": 0.454,
+      "step": 13010
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.6366985998526157e-07,
+      "loss": 0.2401,
+      "step": 13020
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.6352247605011057e-07,
+      "loss": 0.2519,
+      "step": 13030
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.6337509211495947e-07,
+      "loss": 0.2711,
+      "step": 13040
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.632277081798084e-07,
+      "loss": 0.1153,
+      "step": 13050
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.630803242446573e-07,
+      "loss": 0.0948,
+      "step": 13060
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.6293294030950627e-07,
+      "loss": 0.1944,
+      "step": 13070
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.627855563743552e-07,
+      "loss": 0.0131,
+      "step": 13080
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.626381724392041e-07,
+      "loss": 0.1603,
+      "step": 13090
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.6249078850405307e-07,
+      "loss": 0.0004,
+      "step": 13100
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.62343404568902e-07,
+      "loss": 0.0979,
+      "step": 13110
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.621960206337509e-07,
+      "loss": 0.1274,
+      "step": 13120
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.6204863669859986e-07,
+      "loss": 0.0374,
+      "step": 13130
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.619012527634488e-07,
+      "loss": 0.0004,
+      "step": 13140
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.617538688282977e-07,
+      "loss": 0.0031,
+      "step": 13150
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.6160648489314666e-07,
+      "loss": 0.0028,
+      "step": 13160
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.614591009579956e-07,
+      "loss": 0.0003,
+      "step": 13170
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.613117170228445e-07,
+      "loss": 0.7167,
+      "step": 13180
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.6116433308769346e-07,
+      "loss": 0.1597,
+      "step": 13190
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.6101694915254236e-07,
+      "loss": 0.2687,
+      "step": 13200
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.608695652173913e-07,
+      "loss": 0.4846,
+      "step": 13210
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.6072218128224026e-07,
+      "loss": 0.1406,
+      "step": 13220
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.6057479734708916e-07,
+      "loss": 0.1773,
+      "step": 13230
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.604274134119381e-07,
+      "loss": 0.0004,
+      "step": 13240
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.6028002947678706e-07,
+      "loss": 0.0005,
+      "step": 13250
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.6013264554163595e-07,
+      "loss": 0.0003,
+      "step": 13260
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.599852616064849e-07,
+      "loss": 0.0003,
+      "step": 13270
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.5983787767133386e-07,
+      "loss": 0.1207,
+      "step": 13280
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.5969049373618275e-07,
+      "loss": 0.5711,
+      "step": 13290
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.595431098010317e-07,
+      "loss": 0.0119,
+      "step": 13300
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.5939572586588065e-07,
+      "loss": 0.0624,
+      "step": 13310
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.5924834193072955e-07,
+      "loss": 0.1727,
+      "step": 13320
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.591009579955785e-07,
+      "loss": 0.2706,
+      "step": 13330
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.589535740604274e-07,
+      "loss": 0.1166,
+      "step": 13340
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.5880619012527635e-07,
+      "loss": 0.1334,
+      "step": 13350
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.586588061901253e-07,
+      "loss": 0.0004,
+      "step": 13360
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.585114222549742e-07,
+      "loss": 0.2899,
+      "step": 13370
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.5836403831982315e-07,
+      "loss": 0.3726,
+      "step": 13380
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.582166543846721e-07,
+      "loss": 0.1685,
+      "step": 13390
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.58069270449521e-07,
+      "loss": 0.0005,
+      "step": 13400
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.5792188651436995e-07,
+      "loss": 0.2376,
+      "step": 13410
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.577745025792189e-07,
+      "loss": 0.2465,
+      "step": 13420
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.576271186440678e-07,
+      "loss": 0.4121,
+      "step": 13430
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.5747973470891674e-07,
+      "loss": 0.0123,
+      "step": 13440
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.5733235077376564e-07,
+      "loss": 0.0022,
+      "step": 13450
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.571849668386146e-07,
+      "loss": 0.0012,
+      "step": 13460
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.5703758290346354e-07,
+      "loss": 0.266,
+      "step": 13470
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.5689019896831244e-07,
+      "loss": 0.2935,
+      "step": 13480
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.567428150331614e-07,
+      "loss": 0.0043,
+      "step": 13490
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.5659543109801034e-07,
+      "loss": 0.1537,
+      "step": 13500
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.5644804716285924e-07,
+      "loss": 0.0209,
+      "step": 13510
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.563006632277082e-07,
+      "loss": 0.0052,
+      "step": 13520
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.5615327929255714e-07,
+      "loss": 0.1556,
+      "step": 13530
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.5600589535740604e-07,
+      "loss": 0.0345,
+      "step": 13540
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.55858511422255e-07,
+      "loss": 0.2017,
+      "step": 13550
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.5571112748710394e-07,
+      "loss": 0.3678,
+      "step": 13560
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.5556374355195283e-07,
+      "loss": 0.4436,
+      "step": 13570
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.554163596168018e-07,
+      "loss": 0.1847,
+      "step": 13580
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.552689756816507e-07,
+      "loss": 0.0021,
+      "step": 13590
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.5512159174649963e-07,
+      "loss": 0.3647,
+      "step": 13600
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.549742078113486e-07,
+      "loss": 0.001,
+      "step": 13610
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.548268238761975e-07,
+      "loss": 0.2443,
+      "step": 13620
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.5467943994104643e-07,
+      "loss": 0.0907,
+      "step": 13630
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.545320560058954e-07,
+      "loss": 0.007,
+      "step": 13640
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.543846720707443e-07,
+      "loss": 0.2914,
+      "step": 13650
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.5423728813559323e-07,
+      "loss": 0.0518,
+      "step": 13660
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.540899042004422e-07,
+      "loss": 0.1901,
+      "step": 13670
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.539425202652911e-07,
+      "loss": 0.1035,
+      "step": 13680
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.5379513633014003e-07,
+      "loss": 0.3798,
+      "step": 13690
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.53647752394989e-07,
+      "loss": 0.0105,
+      "step": 13700
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.535003684598379e-07,
+      "loss": 0.0345,
+      "step": 13710
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.533529845246868e-07,
+      "loss": 0.0038,
+      "step": 13720
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.532056005895357e-07,
+      "loss": 0.3622,
+      "step": 13730
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.530582166543847e-07,
+      "loss": 0.3421,
+      "step": 13740
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.529108327192336e-07,
+      "loss": 0.0011,
+      "step": 13750
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.527634487840825e-07,
+      "loss": 0.0003,
+      "step": 13760
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.5261606484893147e-07,
+      "loss": 0.2089,
+      "step": 13770
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.524686809137804e-07,
+      "loss": 0.0006,
+      "step": 13780
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.523212969786293e-07,
+      "loss": 0.1881,
+      "step": 13790
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.5217391304347827e-07,
+      "loss": 0.1175,
+      "step": 13800
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.520265291083272e-07,
+      "loss": 0.521,
+      "step": 13810
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.518791451731761e-07,
+      "loss": 0.025,
+      "step": 13820
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.5173176123802507e-07,
+      "loss": 0.1656,
+      "step": 13830
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.5158437730287397e-07,
+      "loss": 0.2863,
+      "step": 13840
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.514369933677229e-07,
+      "loss": 0.4789,
+      "step": 13850
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.5128960943257187e-07,
+      "loss": 0.3231,
+      "step": 13860
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.5114222549742076e-07,
+      "loss": 0.1366,
+      "step": 13870
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.509948415622697e-07,
+      "loss": 0.217,
+      "step": 13880
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.5084745762711866e-07,
+      "loss": 0.163,
+      "step": 13890
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.5070007369196756e-07,
+      "loss": 0.0583,
+      "step": 13900
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.505526897568165e-07,
+      "loss": 0.0943,
+      "step": 13910
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.5040530582166546e-07,
+      "loss": 0.2143,
+      "step": 13920
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.5025792188651436e-07,
+      "loss": 0.2502,
+      "step": 13930
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.501105379513633e-07,
+      "loss": 0.4503,
+      "step": 13940
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.4996315401621226e-07,
+      "loss": 0.0009,
+      "step": 13950
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.4981577008106116e-07,
+      "loss": 0.0973,
+      "step": 13960
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.496683861459101e-07,
+      "loss": 0.0075,
+      "step": 13970
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.49521002210759e-07,
+      "loss": 0.3507,
+      "step": 13980
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.4937361827560796e-07,
+      "loss": 0.366,
+      "step": 13990
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.492262343404569e-07,
+      "loss": 0.0011,
+      "step": 14000
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.490788504053058e-07,
+      "loss": 0.0743,
+      "step": 14010
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.4893146647015476e-07,
+      "loss": 0.342,
+      "step": 14020
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.487840825350037e-07,
+      "loss": 0.0004,
+      "step": 14030
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.486366985998526e-07,
+      "loss": 0.1983,
+      "step": 14040
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.4848931466470155e-07,
+      "loss": 0.2851,
+      "step": 14050
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.483419307295505e-07,
+      "loss": 0.3578,
+      "step": 14060
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.481945467943994e-07,
+      "loss": 0.0005,
+      "step": 14070
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.4804716285924835e-07,
+      "loss": 0.0012,
+      "step": 14080
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.4789977892409725e-07,
+      "loss": 0.1588,
+      "step": 14090
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.477523949889462e-07,
+      "loss": 0.4895,
+      "step": 14100
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.4760501105379515e-07,
+      "loss": 0.1236,
+      "step": 14110
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.4745762711864405e-07,
+      "loss": 0.2594,
+      "step": 14120
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.47310243183493e-07,
+      "loss": 0.1346,
+      "step": 14130
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.4716285924834195e-07,
+      "loss": 0.3469,
+      "step": 14140
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.4701547531319085e-07,
+      "loss": 0.001,
+      "step": 14150
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.468680913780398e-07,
+      "loss": 0.0228,
+      "step": 14160
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.4672070744288875e-07,
+      "loss": 0.0403,
+      "step": 14170
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.4657332350773764e-07,
+      "loss": 0.084,
+      "step": 14180
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.464259395725866e-07,
+      "loss": 0.1499,
+      "step": 14190
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.4627855563743554e-07,
+      "loss": 0.0003,
+      "step": 14200
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.4613117170228444e-07,
+      "loss": 0.6228,
+      "step": 14210
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.459837877671334e-07,
+      "loss": 0.1144,
+      "step": 14220
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.458364038319823e-07,
+      "loss": 0.1836,
+      "step": 14230
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.4568901989683124e-07,
+      "loss": 0.3689,
+      "step": 14240
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.455416359616802e-07,
+      "loss": 0.1652,
+      "step": 14250
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.453942520265291e-07,
+      "loss": 0.0003,
+      "step": 14260
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.4524686809137804e-07,
+      "loss": 0.3052,
+      "step": 14270
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.45099484156227e-07,
+      "loss": 0.0783,
+      "step": 14280
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.449521002210759e-07,
+      "loss": 0.0004,
+      "step": 14290
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.4480471628592484e-07,
+      "loss": 0.0072,
+      "step": 14300
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.446573323507738e-07,
+      "loss": 0.0973,
+      "step": 14310
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.445099484156227e-07,
+      "loss": 0.0008,
+      "step": 14320
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.4436256448047164e-07,
+      "loss": 0.1079,
+      "step": 14330
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.442151805453206e-07,
+      "loss": 0.0012,
+      "step": 14340
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.440677966101695e-07,
+      "loss": 0.2966,
+      "step": 14350
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.4392041267501843e-07,
+      "loss": 0.1403,
+      "step": 14360
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.4377302873986733e-07,
+      "loss": 0.0087,
+      "step": 14370
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.436256448047163e-07,
+      "loss": 0.0408,
+      "step": 14380
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.4347826086956523e-07,
+      "loss": 0.2712,
+      "step": 14390
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.4333087693441413e-07,
+      "loss": 0.1826,
+      "step": 14400
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.431834929992631e-07,
+      "loss": 0.2878,
+      "step": 14410
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.4303610906411203e-07,
+      "loss": 0.1389,
+      "step": 14420
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.4288872512896093e-07,
+      "loss": 0.1586,
+      "step": 14430
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.427413411938099e-07,
+      "loss": 0.1576,
+      "step": 14440
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.4259395725865883e-07,
+      "loss": 0.1944,
+      "step": 14450
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.424465733235077e-07,
+      "loss": 0.084,
+      "step": 14460
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.422991893883567e-07,
+      "loss": 0.0017,
+      "step": 14470
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.421518054532056e-07,
+      "loss": 0.0228,
+      "step": 14480
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.420044215180545e-07,
+      "loss": 0.0006,
+      "step": 14490
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.418570375829035e-07,
+      "loss": 0.1252,
+      "step": 14500
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.4170965364775237e-07,
+      "loss": 0.15,
+      "step": 14510
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.415622697126013e-07,
+      "loss": 0.1083,
+      "step": 14520
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.4141488577745027e-07,
+      "loss": 0.1175,
+      "step": 14530
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.4126750184229917e-07,
+      "loss": 0.6315,
+      "step": 14540
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.411201179071481e-07,
+      "loss": 0.0007,
+      "step": 14550
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.4097273397199707e-07,
+      "loss": 0.0033,
+      "step": 14560
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.4082535003684597e-07,
+      "loss": 0.0002,
+      "step": 14570
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.406779661016949e-07,
+      "loss": 0.3097,
+      "step": 14580
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.4053058216654387e-07,
+      "loss": 0.1803,
+      "step": 14590
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.4038319823139277e-07,
+      "loss": 0.0004,
+      "step": 14600
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.402358142962417e-07,
+      "loss": 0.1375,
+      "step": 14610
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.400884303610906e-07,
+      "loss": 0.3776,
+      "step": 14620
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.3994104642593956e-07,
+      "loss": 0.2265,
+      "step": 14630
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.397936624907885e-07,
+      "loss": 0.1745,
+      "step": 14640
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.396462785556374e-07,
+      "loss": 0.0021,
+      "step": 14650
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.3949889462048636e-07,
+      "loss": 0.0766,
+      "step": 14660
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.393515106853353e-07,
+      "loss": 0.0004,
+      "step": 14670
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.392041267501842e-07,
+      "loss": 0.1751,
+      "step": 14680
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.3905674281503316e-07,
+      "loss": 0.0025,
+      "step": 14690
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.389093588798821e-07,
+      "loss": 0.0081,
+      "step": 14700
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.38761974944731e-07,
+      "loss": 0.249,
+      "step": 14710
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.3861459100957996e-07,
+      "loss": 0.3377,
+      "step": 14720
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.384672070744289e-07,
+      "loss": 0.1793,
+      "step": 14730
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.383198231392778e-07,
+      "loss": 0.002,
+      "step": 14740
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.3817243920412676e-07,
+      "loss": 0.3082,
+      "step": 14750
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.3802505526897566e-07,
+      "loss": 0.0002,
+      "step": 14760
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.378776713338246e-07,
+      "loss": 0.3186,
+      "step": 14770
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.3773028739867356e-07,
+      "loss": 0.1676,
+      "step": 14780
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.3758290346352245e-07,
+      "loss": 0.1601,
+      "step": 14790
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.374355195283714e-07,
+      "loss": 0.1309,
+      "step": 14800
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.3728813559322035e-07,
+      "loss": 0.2299,
+      "step": 14810
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.3714075165806925e-07,
+      "loss": 0.0007,
+      "step": 14820
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.369933677229182e-07,
+      "loss": 0.1454,
+      "step": 14830
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.3684598378776715e-07,
+      "loss": 0.0002,
+      "step": 14840
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.3669859985261605e-07,
+      "loss": 0.1591,
+      "step": 14850
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.36551215917465e-07,
+      "loss": 0.2607,
+      "step": 14860
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.364038319823139e-07,
+      "loss": 0.3347,
+      "step": 14870
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.3625644804716285e-07,
+      "loss": 0.1609,
+      "step": 14880
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.361090641120118e-07,
+      "loss": 0.5976,
+      "step": 14890
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.359616801768607e-07,
+      "loss": 0.0003,
+      "step": 14900
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.3581429624170965e-07,
+      "loss": 0.0003,
+      "step": 14910
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.356669123065586e-07,
+      "loss": 0.1205,
+      "step": 14920
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.355195283714075e-07,
+      "loss": 0.1963,
+      "step": 14930
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.3537214443625644e-07,
+      "loss": 0.166,
+      "step": 14940
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.352247605011054e-07,
+      "loss": 0.0774,
+      "step": 14950
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.350773765659543e-07,
+      "loss": 0.4051,
+      "step": 14960
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.3492999263080324e-07,
+      "loss": 0.0898,
+      "step": 14970
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.347826086956522e-07,
+      "loss": 0.0006,
+      "step": 14980
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.346352247605011e-07,
+      "loss": 0.0033,
+      "step": 14990
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.3448784082535004e-07,
+      "loss": 0.0127,
+      "step": 15000
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.3434045689019894e-07,
+      "loss": 0.1451,
+      "step": 15010
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.341930729550479e-07,
+      "loss": 0.0003,
+      "step": 15020
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.3404568901989684e-07,
+      "loss": 0.2563,
+      "step": 15030
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.3389830508474574e-07,
+      "loss": 0.795,
+      "step": 15040
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.337509211495947e-07,
+      "loss": 0.0368,
+      "step": 15050
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.3360353721444364e-07,
+      "loss": 0.2057,
+      "step": 15060
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.3345615327929254e-07,
+      "loss": 0.1363,
+      "step": 15070
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.333087693441415e-07,
+      "loss": 0.1713,
+      "step": 15080
+    },
+    {
+      "epoch": 5.07,
+      "eval_accuracy": 0.8914151313210505,
+      "eval_loss": 0.6445865631103516,
+      "eval_runtime": 672.5297,
+      "eval_samples_per_second": 3.793,
+      "eval_steps_per_second": 0.949,
+      "step": 15084
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.3316138540899044e-07,
+      "loss": 0.0005,
+      "step": 15090
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.3301400147383933e-07,
+      "loss": 0.0003,
+      "step": 15100
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.328666175386883e-07,
+      "loss": 0.0261,
+      "step": 15110
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.3271923360353723e-07,
+      "loss": 0.0041,
+      "step": 15120
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.3257184966838613e-07,
+      "loss": 0.0009,
+      "step": 15130
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.324244657332351e-07,
+      "loss": 0.0561,
+      "step": 15140
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.32277081798084e-07,
+      "loss": 0.1574,
+      "step": 15150
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.3212969786293293e-07,
+      "loss": 0.0012,
+      "step": 15160
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.319823139277819e-07,
+      "loss": 0.1376,
+      "step": 15170
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.318349299926308e-07,
+      "loss": 0.1453,
+      "step": 15180
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.3168754605747973e-07,
+      "loss": 0.0003,
+      "step": 15190
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.315401621223287e-07,
+      "loss": 0.1791,
+      "step": 15200
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.313927781871776e-07,
+      "loss": 0.4466,
+      "step": 15210
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.312453942520265e-07,
+      "loss": 0.2671,
+      "step": 15220
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.310980103168755e-07,
+      "loss": 0.17,
+      "step": 15230
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.309506263817244e-07,
+      "loss": 0.0002,
+      "step": 15240
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.308032424465733e-07,
+      "loss": 0.3117,
+      "step": 15250
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.306558585114222e-07,
+      "loss": 0.1924,
+      "step": 15260
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.3050847457627117e-07,
+      "loss": 0.3583,
+      "step": 15270
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.303610906411201e-07,
+      "loss": 0.0004,
+      "step": 15280
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.30213706705969e-07,
+      "loss": 0.019,
+      "step": 15290
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.3006632277081797e-07,
+      "loss": 0.3215,
+      "step": 15300
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.299189388356669e-07,
+      "loss": 0.1254,
+      "step": 15310
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.297715549005158e-07,
+      "loss": 0.1912,
+      "step": 15320
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.2962417096536477e-07,
+      "loss": 0.1651,
+      "step": 15330
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.294767870302137e-07,
+      "loss": 0.3174,
+      "step": 15340
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.293294030950626e-07,
+      "loss": 0.1768,
+      "step": 15350
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.2918201915991157e-07,
+      "loss": 0.1098,
+      "step": 15360
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.290346352247605e-07,
+      "loss": 0.1864,
+      "step": 15370
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.288872512896094e-07,
+      "loss": 0.1248,
+      "step": 15380
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.2873986735445837e-07,
+      "loss": 0.1679,
+      "step": 15390
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.2859248341930726e-07,
+      "loss": 0.088,
+      "step": 15400
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.284450994841562e-07,
+      "loss": 0.2566,
+      "step": 15410
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.2829771554900516e-07,
+      "loss": 0.1684,
+      "step": 15420
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.2815033161385406e-07,
+      "loss": 0.2098,
+      "step": 15430
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.28002947678703e-07,
+      "loss": 0.1616,
+      "step": 15440
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.2785556374355196e-07,
+      "loss": 0.0003,
+      "step": 15450
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.2770817980840086e-07,
+      "loss": 0.0943,
+      "step": 15460
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.275607958732498e-07,
+      "loss": 0.0006,
+      "step": 15470
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.2741341193809876e-07,
+      "loss": 0.0103,
+      "step": 15480
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.2726602800294766e-07,
+      "loss": 0.1126,
+      "step": 15490
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.271186440677966e-07,
+      "loss": 0.1907,
+      "step": 15500
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.2697126013264556e-07,
+      "loss": 0.0009,
+      "step": 15510
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.2682387619749446e-07,
+      "loss": 0.1871,
+      "step": 15520
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.266764922623434e-07,
+      "loss": 0.3227,
+      "step": 15530
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.265291083271923e-07,
+      "loss": 0.4686,
+      "step": 15540
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.2638172439204125e-07,
+      "loss": 0.1756,
+      "step": 15550
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.262343404568902e-07,
+      "loss": 0.0022,
+      "step": 15560
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.260869565217391e-07,
+      "loss": 0.0715,
+      "step": 15570
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.2593957258658805e-07,
+      "loss": 0.0004,
+      "step": 15580
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.25792188651437e-07,
+      "loss": 0.0488,
+      "step": 15590
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.256448047162859e-07,
+      "loss": 0.3932,
+      "step": 15600
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.2549742078113485e-07,
+      "loss": 0.0044,
+      "step": 15610
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.253500368459838e-07,
+      "loss": 0.3015,
+      "step": 15620
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.252026529108327e-07,
+      "loss": 0.5223,
+      "step": 15630
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.2505526897568165e-07,
+      "loss": 0.183,
+      "step": 15640
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.2490788504053055e-07,
+      "loss": 0.0016,
+      "step": 15650
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.247605011053795e-07,
+      "loss": 0.0005,
+      "step": 15660
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.2461311717022845e-07,
+      "loss": 0.1088,
+      "step": 15670
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.2446573323507734e-07,
+      "loss": 0.1966,
+      "step": 15680
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.243183492999263e-07,
+      "loss": 0.1602,
+      "step": 15690
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.2417096536477525e-07,
+      "loss": 0.0004,
+      "step": 15700
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.2402358142962414e-07,
+      "loss": 0.2552,
+      "step": 15710
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.238761974944731e-07,
+      "loss": 0.0025,
+      "step": 15720
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.2372881355932204e-07,
+      "loss": 0.4557,
+      "step": 15730
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.2358142962417094e-07,
+      "loss": 0.0005,
+      "step": 15740
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.234340456890199e-07,
+      "loss": 0.0712,
+      "step": 15750
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.2328666175386884e-07,
+      "loss": 0.1603,
+      "step": 15760
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.2313927781871774e-07,
+      "loss": 0.0008,
+      "step": 15770
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.229918938835667e-07,
+      "loss": 0.0006,
+      "step": 15780
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.228445099484156e-07,
+      "loss": 0.0003,
+      "step": 15790
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.2269712601326454e-07,
+      "loss": 0.0027,
+      "step": 15800
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.225497420781135e-07,
+      "loss": 0.4884,
+      "step": 15810
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.224023581429624e-07,
+      "loss": 0.0041,
+      "step": 15820
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.2225497420781134e-07,
+      "loss": 0.0534,
+      "step": 15830
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.221075902726603e-07,
+      "loss": 0.022,
+      "step": 15840
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.219602063375092e-07,
+      "loss": 0.4092,
+      "step": 15850
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.2181282240235813e-07,
+      "loss": 0.0732,
+      "step": 15860
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.216654384672071e-07,
+      "loss": 0.2164,
+      "step": 15870
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.21518054532056e-07,
+      "loss": 0.3203,
+      "step": 15880
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.2137067059690493e-07,
+      "loss": 0.0008,
+      "step": 15890
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.212232866617539e-07,
+      "loss": 0.0672,
+      "step": 15900
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.210759027266028e-07,
+      "loss": 0.3129,
+      "step": 15910
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.2092851879145173e-07,
+      "loss": 0.1704,
+      "step": 15920
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.2078113485630063e-07,
+      "loss": 0.0447,
+      "step": 15930
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.206337509211496e-07,
+      "loss": 0.0013,
+      "step": 15940
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.2048636698599853e-07,
+      "loss": 0.0225,
+      "step": 15950
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.203389830508474e-07,
+      "loss": 0.1793,
+      "step": 15960
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.201915991156964e-07,
+      "loss": 0.0005,
+      "step": 15970
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.2004421518054533e-07,
+      "loss": 0.0007,
+      "step": 15980
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.198968312453942e-07,
+      "loss": 0.0007,
+      "step": 15990
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.197494473102432e-07,
+      "loss": 0.0878,
+      "step": 16000
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.196020633750921e-07,
+      "loss": 0.2037,
+      "step": 16010
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.19454679439941e-07,
+      "loss": 0.0571,
+      "step": 16020
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.1930729550478997e-07,
+      "loss": 0.0472,
+      "step": 16030
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.1915991156963887e-07,
+      "loss": 0.1227,
+      "step": 16040
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.190125276344878e-07,
+      "loss": 0.2009,
+      "step": 16050
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.1886514369933677e-07,
+      "loss": 0.1204,
+      "step": 16060
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.1871775976418567e-07,
+      "loss": 0.0014,
+      "step": 16070
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.185703758290346e-07,
+      "loss": 0.0392,
+      "step": 16080
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.1842299189388357e-07,
+      "loss": 0.0007,
+      "step": 16090
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.1827560795873247e-07,
+      "loss": 0.1108,
+      "step": 16100
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.181282240235814e-07,
+      "loss": 0.3559,
+      "step": 16110
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.1798084008843037e-07,
+      "loss": 0.0003,
+      "step": 16120
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.1783345615327927e-07,
+      "loss": 0.0928,
+      "step": 16130
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.176860722181282e-07,
+      "loss": 0.0881,
+      "step": 16140
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.1753868828297717e-07,
+      "loss": 0.1157,
+      "step": 16150
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.1739130434782606e-07,
+      "loss": 0.2745,
+      "step": 16160
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.17243920412675e-07,
+      "loss": 0.0008,
+      "step": 16170
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.170965364775239e-07,
+      "loss": 0.454,
+      "step": 16180
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.1694915254237286e-07,
+      "loss": 0.0641,
+      "step": 16190
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.168017686072218e-07,
+      "loss": 0.0853,
+      "step": 16200
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.166543846720707e-07,
+      "loss": 0.0995,
+      "step": 16210
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.1650700073691966e-07,
+      "loss": 0.0278,
+      "step": 16220
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.163596168017686e-07,
+      "loss": 0.1673,
+      "step": 16230
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.162122328666175e-07,
+      "loss": 0.1648,
+      "step": 16240
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.1606484893146646e-07,
+      "loss": 0.1665,
+      "step": 16250
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.159174649963154e-07,
+      "loss": 0.0003,
+      "step": 16260
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.157700810611643e-07,
+      "loss": 0.0004,
+      "step": 16270
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.1562269712601326e-07,
+      "loss": 0.0005,
+      "step": 16280
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.1547531319086215e-07,
+      "loss": 0.0002,
+      "step": 16290
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.153279292557111e-07,
+      "loss": 0.1872,
+      "step": 16300
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.1518054532056005e-07,
+      "loss": 0.0012,
+      "step": 16310
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.1503316138540895e-07,
+      "loss": 0.1669,
+      "step": 16320
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.148857774502579e-07,
+      "loss": 0.1629,
+      "step": 16330
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.1473839351510685e-07,
+      "loss": 0.0002,
+      "step": 16340
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.1459100957995575e-07,
+      "loss": 0.1723,
+      "step": 16350
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.144436256448047e-07,
+      "loss": 0.2234,
+      "step": 16360
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.1429624170965365e-07,
+      "loss": 0.0002,
+      "step": 16370
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.1414885777450255e-07,
+      "loss": 0.2757,
+      "step": 16380
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.140014738393515e-07,
+      "loss": 0.1003,
+      "step": 16390
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.1385408990420045e-07,
+      "loss": 0.1771,
+      "step": 16400
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.1370670596904935e-07,
+      "loss": 0.5299,
+      "step": 16410
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.135593220338983e-07,
+      "loss": 0.1982,
+      "step": 16420
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.134119380987472e-07,
+      "loss": 0.3821,
+      "step": 16430
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.1326455416359615e-07,
+      "loss": 0.2827,
+      "step": 16440
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.131171702284451e-07,
+      "loss": 0.4144,
+      "step": 16450
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.12969786293294e-07,
+      "loss": 0.1896,
+      "step": 16460
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.1282240235814294e-07,
+      "loss": 0.0011,
+      "step": 16470
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.126750184229919e-07,
+      "loss": 0.0075,
+      "step": 16480
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.125276344878408e-07,
+      "loss": 0.1475,
+      "step": 16490
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.1238025055268974e-07,
+      "loss": 0.0003,
+      "step": 16500
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.122328666175387e-07,
+      "loss": 0.0001,
+      "step": 16510
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.120854826823876e-07,
+      "loss": 0.2633,
+      "step": 16520
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.1193809874723654e-07,
+      "loss": 0.0004,
+      "step": 16530
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.117907148120855e-07,
+      "loss": 0.1999,
+      "step": 16540
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.116433308769344e-07,
+      "loss": 0.4987,
+      "step": 16550
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.1149594694178334e-07,
+      "loss": 0.0209,
+      "step": 16560
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.1134856300663224e-07,
+      "loss": 0.0003,
+      "step": 16570
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.112011790714812e-07,
+      "loss": 0.0613,
+      "step": 16580
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.1105379513633014e-07,
+      "loss": 0.0108,
+      "step": 16590
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.1090641120117903e-07,
+      "loss": 0.0067,
+      "step": 16600
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.10759027266028e-07,
+      "loss": 0.0001,
+      "step": 16610
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.1061164333087693e-07,
+      "loss": 0.0003,
+      "step": 16620
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.1046425939572583e-07,
+      "loss": 0.0161,
+      "step": 16630
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.103168754605748e-07,
+      "loss": 0.0018,
+      "step": 16640
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.1016949152542373e-07,
+      "loss": 0.392,
+      "step": 16650
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.1002210759027263e-07,
+      "loss": 0.0198,
+      "step": 16660
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.098747236551216e-07,
+      "loss": 0.3065,
+      "step": 16670
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.097273397199705e-07,
+      "loss": 0.2876,
+      "step": 16680
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.0957995578481943e-07,
+      "loss": 0.0003,
+      "step": 16690
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.094325718496684e-07,
+      "loss": 0.3908,
+      "step": 16700
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.092851879145173e-07,
+      "loss": 0.3494,
+      "step": 16710
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.0913780397936623e-07,
+      "loss": 0.2877,
+      "step": 16720
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.089904200442152e-07,
+      "loss": 0.0437,
+      "step": 16730
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.088430361090641e-07,
+      "loss": 0.0609,
+      "step": 16740
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.08695652173913e-07,
+      "loss": 0.1809,
+      "step": 16750
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.08548268238762e-07,
+      "loss": 0.1422,
+      "step": 16760
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.0840088430361087e-07,
+      "loss": 0.0015,
+      "step": 16770
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.082535003684598e-07,
+      "loss": 0.1296,
+      "step": 16780
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.081061164333088e-07,
+      "loss": 0.1274,
+      "step": 16790
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.0795873249815767e-07,
+      "loss": 0.1518,
+      "step": 16800
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.078113485630066e-07,
+      "loss": 0.1166,
+      "step": 16810
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.076639646278555e-07,
+      "loss": 0.3054,
+      "step": 16820
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.0751658069270447e-07,
+      "loss": 0.0001,
+      "step": 16830
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.073691967575534e-07,
+      "loss": 0.1329,
+      "step": 16840
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.072218128224023e-07,
+      "loss": 0.0002,
+      "step": 16850
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.0707442888725127e-07,
+      "loss": 0.0078,
+      "step": 16860
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.069270449521002e-07,
+      "loss": 0.0182,
+      "step": 16870
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.067796610169491e-07,
+      "loss": 0.4244,
+      "step": 16880
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.0663227708179807e-07,
+      "loss": 0.1532,
+      "step": 16890
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.06484893146647e-07,
+      "loss": 0.0101,
+      "step": 16900
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.063375092114959e-07,
+      "loss": 0.0002,
+      "step": 16910
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.0619012527634486e-07,
+      "loss": 0.1771,
+      "step": 16920
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.060427413411938e-07,
+      "loss": 0.0817,
+      "step": 16930
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.058953574060427e-07,
+      "loss": 0.297,
+      "step": 16940
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.0574797347089166e-07,
+      "loss": 0.1162,
+      "step": 16950
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.0560058953574056e-07,
+      "loss": 0.0002,
+      "step": 16960
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.054532056005895e-07,
+      "loss": 0.0004,
+      "step": 16970
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.0530582166543846e-07,
+      "loss": 0.0851,
+      "step": 16980
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.0515843773028736e-07,
+      "loss": 0.0029,
+      "step": 16990
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.050110537951363e-07,
+      "loss": 0.37,
+      "step": 17000
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.0486366985998526e-07,
+      "loss": 0.2724,
+      "step": 17010
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.0471628592483416e-07,
+      "loss": 0.0009,
+      "step": 17020
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.045689019896831e-07,
+      "loss": 0.1432,
+      "step": 17030
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.0442151805453206e-07,
+      "loss": 0.4612,
+      "step": 17040
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.0427413411938095e-07,
+      "loss": 0.0945,
+      "step": 17050
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.041267501842299e-07,
+      "loss": 0.0596,
+      "step": 17060
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.039793662490788e-07,
+      "loss": 0.2178,
+      "step": 17070
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.0383198231392775e-07,
+      "loss": 0.0978,
+      "step": 17080
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.036845983787767e-07,
+      "loss": 0.1647,
+      "step": 17090
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.035372144436256e-07,
+      "loss": 0.1839,
+      "step": 17100
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.0338983050847455e-07,
+      "loss": 0.0152,
+      "step": 17110
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.032424465733235e-07,
+      "loss": 0.2161,
+      "step": 17120
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.030950626381724e-07,
+      "loss": 0.3173,
+      "step": 17130
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.0294767870302135e-07,
+      "loss": 0.2746,
+      "step": 17140
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.028002947678703e-07,
+      "loss": 0.0008,
+      "step": 17150
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.026529108327192e-07,
+      "loss": 0.0017,
+      "step": 17160
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.0250552689756815e-07,
+      "loss": 0.0212,
+      "step": 17170
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.023581429624171e-07,
+      "loss": 0.1723,
+      "step": 17180
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.02210759027266e-07,
+      "loss": 0.0011,
+      "step": 17190
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.0206337509211495e-07,
+      "loss": 0.3933,
+      "step": 17200
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.0191599115696384e-07,
+      "loss": 0.0759,
+      "step": 17210
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.017686072218128e-07,
+      "loss": 0.7277,
+      "step": 17220
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.0162122328666174e-07,
+      "loss": 0.0002,
+      "step": 17230
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.0147383935151064e-07,
+      "loss": 0.5149,
+      "step": 17240
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.013264554163596e-07,
+      "loss": 0.2101,
+      "step": 17250
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.0117907148120854e-07,
+      "loss": 0.0003,
+      "step": 17260
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.0103168754605744e-07,
+      "loss": 0.1705,
+      "step": 17270
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.008843036109064e-07,
+      "loss": 0.0002,
+      "step": 17280
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.0073691967575534e-07,
+      "loss": 0.002,
+      "step": 17290
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.0058953574060424e-07,
+      "loss": 0.0003,
+      "step": 17300
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.004421518054532e-07,
+      "loss": 0.1716,
+      "step": 17310
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.0029476787030214e-07,
+      "loss": 0.1382,
+      "step": 17320
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.0014738393515104e-07,
+      "loss": 0.1029,
+      "step": 17330
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3e-07,
+      "loss": 0.0118,
+      "step": 17340
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.998526160648489e-07,
+      "loss": 0.015,
+      "step": 17350
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.9970523212969783e-07,
+      "loss": 0.0015,
+      "step": 17360
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.995578481945468e-07,
+      "loss": 0.0005,
+      "step": 17370
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.994104642593957e-07,
+      "loss": 0.0046,
+      "step": 17380
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.9926308032424463e-07,
+      "loss": 0.3139,
+      "step": 17390
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.991156963890936e-07,
+      "loss": 0.1326,
+      "step": 17400
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.989683124539425e-07,
+      "loss": 0.2585,
+      "step": 17410
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.9882092851879143e-07,
+      "loss": 0.0001,
+      "step": 17420
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.986735445836404e-07,
+      "loss": 0.2134,
+      "step": 17430
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.985261606484893e-07,
+      "loss": 0.0542,
+      "step": 17440
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.9837877671333823e-07,
+      "loss": 0.0003,
+      "step": 17450
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.982313927781871e-07,
+      "loss": 0.1787,
+      "step": 17460
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.980840088430361e-07,
+      "loss": 0.0002,
+      "step": 17470
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.9793662490788503e-07,
+      "loss": 0.1663,
+      "step": 17480
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.977892409727339e-07,
+      "loss": 0.1904,
+      "step": 17490
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.976418570375829e-07,
+      "loss": 0.1551,
+      "step": 17500
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.974944731024318e-07,
+      "loss": 0.0003,
+      "step": 17510
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.973470891672807e-07,
+      "loss": 0.13,
+      "step": 17520
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.9719970523212967e-07,
+      "loss": 0.001,
+      "step": 17530
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 2.970523212969786e-07,
+      "loss": 0.2547,
+      "step": 17540
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 2.969049373618275e-07,
+      "loss": 0.225,
+      "step": 17550
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 2.9675755342667647e-07,
+      "loss": 0.2122,
+      "step": 17560
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 2.966101694915254e-07,
+      "loss": 0.0556,
+      "step": 17570
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 2.964627855563743e-07,
+      "loss": 0.022,
+      "step": 17580
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 2.9631540162122327e-07,
+      "loss": 0.006,
+      "step": 17590
+    },
+    {
+      "epoch": 6.07,
+      "eval_accuracy": 0.8918071344570757,
+      "eval_loss": 0.6782776117324829,
+      "eval_runtime": 670.7184,
+      "eval_samples_per_second": 3.803,
+      "eval_steps_per_second": 0.951,
+      "step": 17598
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.9616801768607217e-07,
+      "loss": 0.6321,
+      "step": 17600
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.960206337509211e-07,
+      "loss": 0.2074,
+      "step": 17610
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.9587324981577007e-07,
+      "loss": 0.0003,
+      "step": 17620
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.9572586588061897e-07,
+      "loss": 0.1349,
+      "step": 17630
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.955784819454679e-07,
+      "loss": 0.0695,
+      "step": 17640
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.9543109801031687e-07,
+      "loss": 0.0002,
+      "step": 17650
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.9528371407516576e-07,
+      "loss": 0.0002,
+      "step": 17660
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.951363301400147e-07,
+      "loss": 0.164,
+      "step": 17670
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.9498894620486366e-07,
+      "loss": 0.0141,
+      "step": 17680
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.9484156226971256e-07,
+      "loss": 0.1056,
+      "step": 17690
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.946941783345615e-07,
+      "loss": 0.0977,
+      "step": 17700
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.9454679439941046e-07,
+      "loss": 0.0002,
+      "step": 17710
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.9439941046425936e-07,
+      "loss": 0.001,
+      "step": 17720
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.942520265291083e-07,
+      "loss": 0.5057,
+      "step": 17730
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.941046425939572e-07,
+      "loss": 0.3537,
+      "step": 17740
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.9395725865880616e-07,
+      "loss": 0.2192,
+      "step": 17750
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.938098747236551e-07,
+      "loss": 0.1215,
+      "step": 17760
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.93662490788504e-07,
+      "loss": 0.058,
+      "step": 17770
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.9351510685335296e-07,
+      "loss": 0.0002,
+      "step": 17780
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.933677229182019e-07,
+      "loss": 0.0001,
+      "step": 17790
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.932203389830508e-07,
+      "loss": 0.0003,
+      "step": 17800
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.9307295504789975e-07,
+      "loss": 0.0463,
+      "step": 17810
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.929255711127487e-07,
+      "loss": 0.139,
+      "step": 17820
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.927781871775976e-07,
+      "loss": 0.0107,
+      "step": 17830
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.9263080324244655e-07,
+      "loss": 0.0541,
+      "step": 17840
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.9248341930729545e-07,
+      "loss": 0.0872,
+      "step": 17850
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.923360353721444e-07,
+      "loss": 0.0002,
+      "step": 17860
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.9218865143699335e-07,
+      "loss": 0.3727,
+      "step": 17870
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.9204126750184225e-07,
+      "loss": 0.0002,
+      "step": 17880
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.918938835666912e-07,
+      "loss": 0.0659,
+      "step": 17890
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.9174649963154015e-07,
+      "loss": 0.1454,
+      "step": 17900
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.9159911569638905e-07,
+      "loss": 0.1001,
+      "step": 17910
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.91451731761238e-07,
+      "loss": 0.0002,
+      "step": 17920
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.9130434782608695e-07,
+      "loss": 0.192,
+      "step": 17930
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.9115696389093585e-07,
+      "loss": 0.0029,
+      "step": 17940
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.910095799557848e-07,
+      "loss": 0.2828,
+      "step": 17950
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.9086219602063375e-07,
+      "loss": 0.0006,
+      "step": 17960
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.9071481208548264e-07,
+      "loss": 0.0013,
+      "step": 17970
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.905674281503316e-07,
+      "loss": 0.1458,
+      "step": 17980
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.904200442151805e-07,
+      "loss": 0.0002,
+      "step": 17990
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.9027266028002944e-07,
+      "loss": 0.166,
+      "step": 18000
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.901252763448784e-07,
+      "loss": 0.0002,
+      "step": 18010
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.899778924097273e-07,
+      "loss": 0.0072,
+      "step": 18020
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.8983050847457624e-07,
+      "loss": 0.1308,
+      "step": 18030
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.896831245394252e-07,
+      "loss": 0.0002,
+      "step": 18040
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.895357406042741e-07,
+      "loss": 0.0409,
+      "step": 18050
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.8938835666912304e-07,
+      "loss": 0.3293,
+      "step": 18060
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.89240972733972e-07,
+      "loss": 0.3794,
+      "step": 18070
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.890935887988209e-07,
+      "loss": 0.0013,
+      "step": 18080
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.8894620486366984e-07,
+      "loss": 0.0004,
+      "step": 18090
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.8879882092851873e-07,
+      "loss": 0.1924,
+      "step": 18100
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.886514369933677e-07,
+      "loss": 0.2518,
+      "step": 18110
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.8850405305821663e-07,
+      "loss": 0.0003,
+      "step": 18120
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.8835666912306553e-07,
+      "loss": 0.0003,
+      "step": 18130
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.882092851879145e-07,
+      "loss": 0.0007,
+      "step": 18140
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.8806190125276343e-07,
+      "loss": 0.0003,
+      "step": 18150
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.8791451731761233e-07,
+      "loss": 0.0768,
+      "step": 18160
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.877671333824613e-07,
+      "loss": 0.0916,
+      "step": 18170
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.8761974944731023e-07,
+      "loss": 0.1593,
+      "step": 18180
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.8747236551215913e-07,
+      "loss": 0.296,
+      "step": 18190
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.873249815770081e-07,
+      "loss": 0.0031,
+      "step": 18200
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.8717759764185703e-07,
+      "loss": 0.1721,
+      "step": 18210
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.8703021370670593e-07,
+      "loss": 0.3909,
+      "step": 18220
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.868828297715549e-07,
+      "loss": 0.0003,
+      "step": 18230
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.867354458364038e-07,
+      "loss": 0.5581,
+      "step": 18240
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.865880619012527e-07,
+      "loss": 0.1555,
+      "step": 18250
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.864406779661017e-07,
+      "loss": 0.097,
+      "step": 18260
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.8629329403095057e-07,
+      "loss": 0.1838,
+      "step": 18270
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.861459100957995e-07,
+      "loss": 0.0008,
+      "step": 18280
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.859985261606485e-07,
+      "loss": 0.1394,
+      "step": 18290
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.8585114222549737e-07,
+      "loss": 0.2273,
+      "step": 18300
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.857037582903463e-07,
+      "loss": 0.0001,
+      "step": 18310
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.8555637435519527e-07,
+      "loss": 0.305,
+      "step": 18320
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.8540899042004417e-07,
+      "loss": 0.1991,
+      "step": 18330
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.852616064848931e-07,
+      "loss": 0.0009,
+      "step": 18340
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.8511422254974207e-07,
+      "loss": 0.0614,
+      "step": 18350
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.8496683861459097e-07,
+      "loss": 0.0002,
+      "step": 18360
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.848194546794399e-07,
+      "loss": 0.0005,
+      "step": 18370
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.846720707442888e-07,
+      "loss": 0.1395,
+      "step": 18380
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.8452468680913777e-07,
+      "loss": 0.1751,
+      "step": 18390
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.843773028739867e-07,
+      "loss": 0.1159,
+      "step": 18400
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.842299189388356e-07,
+      "loss": 0.0003,
+      "step": 18410
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.8408253500368456e-07,
+      "loss": 0.168,
+      "step": 18420
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.839351510685335e-07,
+      "loss": 0.0025,
+      "step": 18430
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.837877671333824e-07,
+      "loss": 0.1314,
+      "step": 18440
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.8364038319823136e-07,
+      "loss": 0.2606,
+      "step": 18450
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.834929992630803e-07,
+      "loss": 0.2697,
+      "step": 18460
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.833456153279292e-07,
+      "loss": 0.2029,
+      "step": 18470
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.8319823139277816e-07,
+      "loss": 0.0566,
+      "step": 18480
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.8305084745762706e-07,
+      "loss": 0.0038,
+      "step": 18490
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.82903463522476e-07,
+      "loss": 0.0082,
+      "step": 18500
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.8275607958732496e-07,
+      "loss": 0.0004,
+      "step": 18510
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.8260869565217386e-07,
+      "loss": 0.1301,
+      "step": 18520
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.824613117170228e-07,
+      "loss": 0.0001,
+      "step": 18530
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.8231392778187176e-07,
+      "loss": 0.0002,
+      "step": 18540
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.8216654384672065e-07,
+      "loss": 0.2452,
+      "step": 18550
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.820191599115696e-07,
+      "loss": 0.1076,
+      "step": 18560
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.8187177597641856e-07,
+      "loss": 0.0003,
+      "step": 18570
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.8172439204126745e-07,
+      "loss": 0.2386,
+      "step": 18580
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.815770081061164e-07,
+      "loss": 0.2018,
+      "step": 18590
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.8142962417096535e-07,
+      "loss": 0.1397,
+      "step": 18600
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.8128224023581425e-07,
+      "loss": 0.0196,
+      "step": 18610
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.811348563006632e-07,
+      "loss": 0.0097,
+      "step": 18620
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.809874723655121e-07,
+      "loss": 0.0248,
+      "step": 18630
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.8084008843036105e-07,
+      "loss": 0.0021,
+      "step": 18640
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.8069270449521e-07,
+      "loss": 0.1038,
+      "step": 18650
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.805453205600589e-07,
+      "loss": 0.0002,
+      "step": 18660
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.8039793662490785e-07,
+      "loss": 0.3384,
+      "step": 18670
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.802505526897568e-07,
+      "loss": 0.057,
+      "step": 18680
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.801031687546057e-07,
+      "loss": 0.0002,
+      "step": 18690
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.7995578481945465e-07,
+      "loss": 0.4211,
+      "step": 18700
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.798084008843036e-07,
+      "loss": 0.015,
+      "step": 18710
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.796610169491525e-07,
+      "loss": 0.2448,
+      "step": 18720
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.7951363301400144e-07,
+      "loss": 0.131,
+      "step": 18730
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.793662490788504e-07,
+      "loss": 0.1073,
+      "step": 18740
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.792188651436993e-07,
+      "loss": 0.12,
+      "step": 18750
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.7907148120854824e-07,
+      "loss": 0.0001,
+      "step": 18760
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.7892409727339714e-07,
+      "loss": 0.0002,
+      "step": 18770
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.787767133382461e-07,
+      "loss": 0.0002,
+      "step": 18780
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.7862932940309504e-07,
+      "loss": 0.1564,
+      "step": 18790
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.7848194546794394e-07,
+      "loss": 0.0821,
+      "step": 18800
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.783345615327929e-07,
+      "loss": 0.1561,
+      "step": 18810
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.7818717759764184e-07,
+      "loss": 0.1846,
+      "step": 18820
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.7803979366249074e-07,
+      "loss": 0.0007,
+      "step": 18830
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.778924097273397e-07,
+      "loss": 0.1214,
+      "step": 18840
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.7774502579218864e-07,
+      "loss": 0.0005,
+      "step": 18850
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.7759764185703753e-07,
+      "loss": 0.0002,
+      "step": 18860
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.774502579218865e-07,
+      "loss": 0.0001,
+      "step": 18870
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.773028739867354e-07,
+      "loss": 0.1048,
+      "step": 18880
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.7715549005158433e-07,
+      "loss": 0.0004,
+      "step": 18890
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.770081061164333e-07,
+      "loss": 0.0001,
+      "step": 18900
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.768607221812822e-07,
+      "loss": 0.0285,
+      "step": 18910
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.7671333824613113e-07,
+      "loss": 0.0002,
+      "step": 18920
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.765659543109801e-07,
+      "loss": 0.2019,
+      "step": 18930
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.76418570375829e-07,
+      "loss": 0.0002,
+      "step": 18940
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.7627118644067793e-07,
+      "loss": 0.2407,
+      "step": 18950
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.761238025055269e-07,
+      "loss": 0.0213,
+      "step": 18960
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.759764185703758e-07,
+      "loss": 0.1889,
+      "step": 18970
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.7582903463522473e-07,
+      "loss": 0.0782,
+      "step": 18980
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.756816507000737e-07,
+      "loss": 0.1622,
+      "step": 18990
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.755342667649226e-07,
+      "loss": 0.2291,
+      "step": 19000
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.753868828297715e-07,
+      "loss": 0.1466,
+      "step": 19010
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.752394988946204e-07,
+      "loss": 0.1118,
+      "step": 19020
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.750921149594694e-07,
+      "loss": 0.1056,
+      "step": 19030
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.749447310243183e-07,
+      "loss": 0.0001,
+      "step": 19040
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.747973470891672e-07,
+      "loss": 0.0096,
+      "step": 19050
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.7464996315401617e-07,
+      "loss": 0.3279,
+      "step": 19060
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.745025792188651e-07,
+      "loss": 0.1436,
+      "step": 19070
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.74355195283714e-07,
+      "loss": 0.3979,
+      "step": 19080
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.7420781134856297e-07,
+      "loss": 0.0003,
+      "step": 19090
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.740604274134119e-07,
+      "loss": 0.167,
+      "step": 19100
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.739130434782608e-07,
+      "loss": 0.1925,
+      "step": 19110
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.7376565954310977e-07,
+      "loss": 0.1341,
+      "step": 19120
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.736182756079587e-07,
+      "loss": 0.0005,
+      "step": 19130
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.734708916728076e-07,
+      "loss": 0.1123,
+      "step": 19140
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.7332350773765657e-07,
+      "loss": 0.2532,
+      "step": 19150
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.7317612380250546e-07,
+      "loss": 0.0003,
+      "step": 19160
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.730287398673544e-07,
+      "loss": 0.2715,
+      "step": 19170
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.7288135593220336e-07,
+      "loss": 0.1016,
+      "step": 19180
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.7273397199705226e-07,
+      "loss": 0.124,
+      "step": 19190
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.7258658806190127e-07,
+      "loss": 0.0001,
+      "step": 19200
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.724392041267502e-07,
+      "loss": 0.4932,
+      "step": 19210
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.722918201915991e-07,
+      "loss": 0.3676,
+      "step": 19220
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.7214443625644806e-07,
+      "loss": 0.3588,
+      "step": 19230
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.71997052321297e-07,
+      "loss": 0.051,
+      "step": 19240
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.718496683861459e-07,
+      "loss": 0.0166,
+      "step": 19250
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.7170228445099486e-07,
+      "loss": 0.3293,
+      "step": 19260
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.715549005158438e-07,
+      "loss": 0.0004,
+      "step": 19270
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.714075165806927e-07,
+      "loss": 0.1444,
+      "step": 19280
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.7126013264554166e-07,
+      "loss": 0.2319,
+      "step": 19290
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.7111274871039056e-07,
+      "loss": 0.0004,
+      "step": 19300
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.709653647752395e-07,
+      "loss": 0.2146,
+      "step": 19310
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.7081798084008846e-07,
+      "loss": 0.0751,
+      "step": 19320
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.7067059690493736e-07,
+      "loss": 0.1525,
+      "step": 19330
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.705232129697863e-07,
+      "loss": 0.1254,
+      "step": 19340
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.7037582903463526e-07,
+      "loss": 0.0005,
+      "step": 19350
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.7022844509948415e-07,
+      "loss": 0.2601,
+      "step": 19360
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.700810611643331e-07,
+      "loss": 0.0076,
+      "step": 19370
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.6993367722918206e-07,
+      "loss": 0.0002,
+      "step": 19380
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.6978629329403095e-07,
+      "loss": 0.0015,
+      "step": 19390
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.696389093588799e-07,
+      "loss": 0.0008,
+      "step": 19400
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.6949152542372885e-07,
+      "loss": 0.3851,
+      "step": 19410
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.6934414148857775e-07,
+      "loss": 0.2189,
+      "step": 19420
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.691967575534267e-07,
+      "loss": 0.0003,
+      "step": 19430
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.690493736182756e-07,
+      "loss": 0.0003,
+      "step": 19440
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.6890198968312455e-07,
+      "loss": 0.0003,
+      "step": 19450
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.687546057479735e-07,
+      "loss": 0.1555,
+      "step": 19460
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.686072218128224e-07,
+      "loss": 0.0001,
+      "step": 19470
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.6845983787767135e-07,
+      "loss": 0.1884,
+      "step": 19480
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.683124539425203e-07,
+      "loss": 0.1458,
+      "step": 19490
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.681650700073692e-07,
+      "loss": 0.1693,
+      "step": 19500
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.6801768607221815e-07,
+      "loss": 0.0066,
+      "step": 19510
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.678703021370671e-07,
+      "loss": 0.0002,
+      "step": 19520
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.67722918201916e-07,
+      "loss": 0.2034,
+      "step": 19530
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.6757553426676494e-07,
+      "loss": 0.2694,
+      "step": 19540
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.6742815033161384e-07,
+      "loss": 0.0065,
+      "step": 19550
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.672807663964628e-07,
+      "loss": 0.0002,
+      "step": 19560
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.6713338246131174e-07,
+      "loss": 0.0004,
+      "step": 19570
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.6698599852616064e-07,
+      "loss": 0.0003,
+      "step": 19580
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.668386145910096e-07,
+      "loss": 0.1862,
+      "step": 19590
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.6669123065585854e-07,
+      "loss": 0.2973,
+      "step": 19600
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.6654384672070744e-07,
+      "loss": 0.2044,
+      "step": 19610
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.663964627855564e-07,
+      "loss": 0.0088,
+      "step": 19620
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.6624907885040534e-07,
+      "loss": 0.0005,
+      "step": 19630
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.6610169491525424e-07,
+      "loss": 0.1317,
+      "step": 19640
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.659543109801032e-07,
+      "loss": 0.0139,
+      "step": 19650
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.6580692704495214e-07,
+      "loss": 0.3263,
+      "step": 19660
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.6565954310980103e-07,
+      "loss": 0.2684,
+      "step": 19670
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.6551215917465e-07,
+      "loss": 0.1531,
+      "step": 19680
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.653647752394989e-07,
+      "loss": 0.004,
+      "step": 19690
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.6521739130434783e-07,
+      "loss": 0.0002,
+      "step": 19700
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.650700073691968e-07,
+      "loss": 0.1341,
+      "step": 19710
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.649226234340457e-07,
+      "loss": 0.0174,
+      "step": 19720
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.6477523949889463e-07,
+      "loss": 0.1536,
+      "step": 19730
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.646278555637436e-07,
+      "loss": 0.0018,
+      "step": 19740
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.644804716285925e-07,
+      "loss": 0.0836,
+      "step": 19750
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.6433308769344143e-07,
+      "loss": 0.0002,
+      "step": 19760
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.641857037582904e-07,
+      "loss": 0.305,
+      "step": 19770
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.640383198231393e-07,
+      "loss": 0.3056,
+      "step": 19780
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.6389093588798823e-07,
+      "loss": 0.1846,
+      "step": 19790
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.637435519528371e-07,
+      "loss": 0.0002,
+      "step": 19800
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.635961680176861e-07,
+      "loss": 0.1785,
+      "step": 19810
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.63448784082535e-07,
+      "loss": 0.2134,
+      "step": 19820
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.633014001473839e-07,
+      "loss": 0.1045,
+      "step": 19830
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.6315401621223287e-07,
+      "loss": 0.0001,
+      "step": 19840
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.630066322770818e-07,
+      "loss": 0.1259,
+      "step": 19850
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.628592483419307e-07,
+      "loss": 0.1095,
+      "step": 19860
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.6271186440677967e-07,
+      "loss": 0.2348,
+      "step": 19870
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.625644804716286e-07,
+      "loss": 0.3533,
+      "step": 19880
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.624170965364775e-07,
+      "loss": 0.0875,
+      "step": 19890
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.6226971260132647e-07,
+      "loss": 0.0002,
+      "step": 19900
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.621223286661754e-07,
+      "loss": 0.3133,
+      "step": 19910
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.619749447310243e-07,
+      "loss": 0.0003,
+      "step": 19920
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.6182756079587327e-07,
+      "loss": 0.1281,
+      "step": 19930
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.6168017686072217e-07,
+      "loss": 0.0009,
+      "step": 19940
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.615327929255711e-07,
+      "loss": 0.3907,
+      "step": 19950
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.6138540899042007e-07,
+      "loss": 0.2036,
+      "step": 19960
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.6123802505526896e-07,
+      "loss": 0.1012,
+      "step": 19970
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.610906411201179e-07,
+      "loss": 0.1102,
+      "step": 19980
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.6094325718496686e-07,
+      "loss": 0.0027,
+      "step": 19990
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.6079587324981576e-07,
+      "loss": 0.2064,
+      "step": 20000
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.606484893146647e-07,
+      "loss": 0.1783,
+      "step": 20010
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.6050110537951366e-07,
+      "loss": 0.2118,
+      "step": 20020
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.6035372144436256e-07,
+      "loss": 0.3706,
+      "step": 20030
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.602063375092115e-07,
+      "loss": 0.0063,
+      "step": 20040
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 2.6005895357406046e-07,
+      "loss": 0.1138,
+      "step": 20050
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 2.5991156963890936e-07,
+      "loss": 0.3372,
+      "step": 20060
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 2.597641857037583e-07,
+      "loss": 0.1012,
+      "step": 20070
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 2.596168017686072e-07,
+      "loss": 0.1983,
+      "step": 20080
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 2.5946941783345616e-07,
+      "loss": 0.3283,
+      "step": 20090
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 2.593220338983051e-07,
+      "loss": 0.3026,
+      "step": 20100
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 2.59174649963154e-07,
+      "loss": 0.004,
+      "step": 20110
+    },
+    {
+      "epoch": 7.07,
+      "eval_accuracy": 0.8863190905527244,
+      "eval_loss": 0.7586002349853516,
+      "eval_runtime": 710.3238,
+      "eval_samples_per_second": 3.591,
+      "eval_steps_per_second": 0.898,
+      "step": 20112
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.5902726602800295e-07,
+      "loss": 0.3788,
+      "step": 20120
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.588798820928519e-07,
+      "loss": 0.0402,
+      "step": 20130
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.587324981577008e-07,
+      "loss": 0.0003,
+      "step": 20140
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.5858511422254975e-07,
+      "loss": 0.0377,
+      "step": 20150
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.584377302873987e-07,
+      "loss": 0.1889,
+      "step": 20160
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.582903463522476e-07,
+      "loss": 0.0006,
+      "step": 20170
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.5814296241709655e-07,
+      "loss": 0.5983,
+      "step": 20180
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.5799557848194545e-07,
+      "loss": 0.2864,
+      "step": 20190
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.578481945467944e-07,
+      "loss": 0.3506,
+      "step": 20200
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.5770081061164335e-07,
+      "loss": 0.0006,
+      "step": 20210
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.5755342667649225e-07,
+      "loss": 0.2777,
+      "step": 20220
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.574060427413412e-07,
+      "loss": 0.6107,
+      "step": 20230
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.5725865880619015e-07,
+      "loss": 0.2825,
+      "step": 20240
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.5711127487103905e-07,
+      "loss": 0.0032,
+      "step": 20250
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.56963890935888e-07,
+      "loss": 0.0004,
+      "step": 20260
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.5681650700073695e-07,
+      "loss": 0.0135,
+      "step": 20270
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.5666912306558584e-07,
+      "loss": 0.0002,
+      "step": 20280
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.565217391304348e-07,
+      "loss": 0.0003,
+      "step": 20290
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.5637435519528374e-07,
+      "loss": 0.0853,
+      "step": 20300
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.5622697126013264e-07,
+      "loss": 0.0007,
+      "step": 20310
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.560795873249816e-07,
+      "loss": 0.1098,
+      "step": 20320
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.559322033898305e-07,
+      "loss": 0.002,
+      "step": 20330
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.5578481945467944e-07,
+      "loss": 0.0006,
+      "step": 20340
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.556374355195284e-07,
+      "loss": 0.1185,
+      "step": 20350
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.554900515843773e-07,
+      "loss": 0.0856,
+      "step": 20360
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.5534266764922624e-07,
+      "loss": 0.3431,
+      "step": 20370
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.551952837140752e-07,
+      "loss": 0.0003,
+      "step": 20380
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.550478997789241e-07,
+      "loss": 0.1533,
+      "step": 20390
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.5490051584377304e-07,
+      "loss": 0.0037,
+      "step": 20400
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.54753131908622e-07,
+      "loss": 0.0001,
+      "step": 20410
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.546057479734709e-07,
+      "loss": 0.0108,
+      "step": 20420
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.5445836403831983e-07,
+      "loss": 0.1847,
+      "step": 20430
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.543109801031688e-07,
+      "loss": 0.0005,
+      "step": 20440
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.541635961680177e-07,
+      "loss": 0.4133,
+      "step": 20450
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.5401621223286663e-07,
+      "loss": 0.0003,
+      "step": 20460
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.5386882829771553e-07,
+      "loss": 0.0001,
+      "step": 20470
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.537214443625645e-07,
+      "loss": 0.0011,
+      "step": 20480
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.5357406042741343e-07,
+      "loss": 0.071,
+      "step": 20490
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.5342667649226233e-07,
+      "loss": 0.0002,
+      "step": 20500
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.532792925571113e-07,
+      "loss": 0.2056,
+      "step": 20510
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.5313190862196023e-07,
+      "loss": 0.0002,
+      "step": 20520
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.5298452468680913e-07,
+      "loss": 0.0001,
+      "step": 20530
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.528371407516581e-07,
+      "loss": 0.0001,
+      "step": 20540
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.5268975681650703e-07,
+      "loss": 0.0003,
+      "step": 20550
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.525423728813559e-07,
+      "loss": 0.0003,
+      "step": 20560
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.523949889462049e-07,
+      "loss": 0.1256,
+      "step": 20570
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.5224760501105377e-07,
+      "loss": 0.0001,
+      "step": 20580
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.521002210759027e-07,
+      "loss": 0.1684,
+      "step": 20590
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.519528371407517e-07,
+      "loss": 0.1931,
+      "step": 20600
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.5180545320560057e-07,
+      "loss": 0.0002,
+      "step": 20610
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.516580692704495e-07,
+      "loss": 0.0001,
+      "step": 20620
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.5151068533529847e-07,
+      "loss": 0.0006,
+      "step": 20630
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.5136330140014737e-07,
+      "loss": 0.1576,
+      "step": 20640
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.512159174649963e-07,
+      "loss": 0.144,
+      "step": 20650
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.5106853352984527e-07,
+      "loss": 0.0005,
+      "step": 20660
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.5092114959469417e-07,
+      "loss": 0.155,
+      "step": 20670
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.507737656595431e-07,
+      "loss": 0.0005,
+      "step": 20680
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.5062638172439207e-07,
+      "loss": 0.1244,
+      "step": 20690
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.5047899778924097e-07,
+      "loss": 0.1394,
+      "step": 20700
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.503316138540899e-07,
+      "loss": 0.1049,
+      "step": 20710
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.501842299189388e-07,
+      "loss": 0.1664,
+      "step": 20720
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.5003684598378776e-07,
+      "loss": 0.0002,
+      "step": 20730
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.498894620486367e-07,
+      "loss": 0.1461,
+      "step": 20740
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.497420781134856e-07,
+      "loss": 0.0002,
+      "step": 20750
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.4959469417833456e-07,
+      "loss": 0.0002,
+      "step": 20760
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.494473102431835e-07,
+      "loss": 0.1396,
+      "step": 20770
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.492999263080324e-07,
+      "loss": 0.1869,
+      "step": 20780
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.4915254237288136e-07,
+      "loss": 0.0059,
+      "step": 20790
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.490051584377303e-07,
+      "loss": 0.0002,
+      "step": 20800
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.488577745025792e-07,
+      "loss": 0.1787,
+      "step": 20810
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.4871039056742816e-07,
+      "loss": 0.1621,
+      "step": 20820
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.485630066322771e-07,
+      "loss": 0.1469,
+      "step": 20830
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.48415622697126e-07,
+      "loss": 0.0002,
+      "step": 20840
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.4826823876197496e-07,
+      "loss": 0.0001,
+      "step": 20850
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.4812085482682385e-07,
+      "loss": 0.1019,
+      "step": 20860
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.479734708916728e-07,
+      "loss": 0.0004,
+      "step": 20870
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.4782608695652176e-07,
+      "loss": 0.1093,
+      "step": 20880
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.4767870302137065e-07,
+      "loss": 0.273,
+      "step": 20890
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.475313190862196e-07,
+      "loss": 0.0236,
+      "step": 20900
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.4738393515106855e-07,
+      "loss": 0.0001,
+      "step": 20910
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.4723655121591745e-07,
+      "loss": 0.2027,
+      "step": 20920
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.470891672807664e-07,
+      "loss": 0.1292,
+      "step": 20930
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.4694178334561535e-07,
+      "loss": 0.0003,
+      "step": 20940
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.4679439941046425e-07,
+      "loss": 0.0262,
+      "step": 20950
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.466470154753132e-07,
+      "loss": 0.0002,
+      "step": 20960
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.464996315401621e-07,
+      "loss": 0.0002,
+      "step": 20970
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.4635224760501105e-07,
+      "loss": 0.059,
+      "step": 20980
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.4620486366986e-07,
+      "loss": 0.0002,
+      "step": 20990
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.460574797347089e-07,
+      "loss": 0.0941,
+      "step": 21000
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.4591009579955785e-07,
+      "loss": 0.1558,
+      "step": 21010
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.457627118644068e-07,
+      "loss": 0.0003,
+      "step": 21020
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.456153279292557e-07,
+      "loss": 0.0003,
+      "step": 21030
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.4546794399410464e-07,
+      "loss": 0.0239,
+      "step": 21040
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.453205600589536e-07,
+      "loss": 0.0003,
+      "step": 21050
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.451731761238025e-07,
+      "loss": 0.0003,
+      "step": 21060
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.4502579218865144e-07,
+      "loss": 0.0016,
+      "step": 21070
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.448784082535004e-07,
+      "loss": 0.0001,
+      "step": 21080
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.447310243183493e-07,
+      "loss": 0.2107,
+      "step": 21090
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.4458364038319824e-07,
+      "loss": 0.0072,
+      "step": 21100
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.4443625644804714e-07,
+      "loss": 0.0001,
+      "step": 21110
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.442888725128961e-07,
+      "loss": 0.0001,
+      "step": 21120
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.4414148857774504e-07,
+      "loss": 0.0001,
+      "step": 21130
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.4399410464259394e-07,
+      "loss": 0.0002,
+      "step": 21140
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.438467207074429e-07,
+      "loss": 0.0043,
+      "step": 21150
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.4369933677229184e-07,
+      "loss": 0.0005,
+      "step": 21160
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.4355195283714073e-07,
+      "loss": 0.1504,
+      "step": 21170
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.434045689019897e-07,
+      "loss": 0.0049,
+      "step": 21180
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.4325718496683864e-07,
+      "loss": 0.0002,
+      "step": 21190
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.4310980103168753e-07,
+      "loss": 0.0001,
+      "step": 21200
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.429624170965365e-07,
+      "loss": 0.2885,
+      "step": 21210
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.4281503316138543e-07,
+      "loss": 0.003,
+      "step": 21220
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.4266764922623433e-07,
+      "loss": 0.17,
+      "step": 21230
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.425202652910833e-07,
+      "loss": 0.0001,
+      "step": 21240
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.423728813559322e-07,
+      "loss": 0.1944,
+      "step": 21250
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.4222549742078113e-07,
+      "loss": 0.1203,
+      "step": 21260
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.420781134856301e-07,
+      "loss": 0.1075,
+      "step": 21270
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.41930729550479e-07,
+      "loss": 0.0007,
+      "step": 21280
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.4178334561532793e-07,
+      "loss": 0.1235,
+      "step": 21290
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.416359616801769e-07,
+      "loss": 0.2031,
+      "step": 21300
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.414885777450258e-07,
+      "loss": 0.0002,
+      "step": 21310
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.413411938098747e-07,
+      "loss": 0.3113,
+      "step": 21320
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.411938098747237e-07,
+      "loss": 0.008,
+      "step": 21330
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.410464259395726e-07,
+      "loss": 0.2122,
+      "step": 21340
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.408990420044215e-07,
+      "loss": 0.1584,
+      "step": 21350
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.407516580692704e-07,
+      "loss": 0.021,
+      "step": 21360
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.4060427413411937e-07,
+      "loss": 0.1273,
+      "step": 21370
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.404568901989683e-07,
+      "loss": 0.1911,
+      "step": 21380
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.403095062638172e-07,
+      "loss": 0.1599,
+      "step": 21390
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.4016212232866617e-07,
+      "loss": 0.0029,
+      "step": 21400
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.400147383935151e-07,
+      "loss": 0.3969,
+      "step": 21410
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.39867354458364e-07,
+      "loss": 0.4632,
+      "step": 21420
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.3971997052321297e-07,
+      "loss": 0.0001,
+      "step": 21430
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.395725865880619e-07,
+      "loss": 0.1942,
+      "step": 21440
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.394252026529108e-07,
+      "loss": 0.0344,
+      "step": 21450
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3927781871775977e-07,
+      "loss": 0.0002,
+      "step": 21460
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.391304347826087e-07,
+      "loss": 0.1698,
+      "step": 21470
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.389830508474576e-07,
+      "loss": 0.1323,
+      "step": 21480
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3883566691230656e-07,
+      "loss": 0.0002,
+      "step": 21490
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3868828297715546e-07,
+      "loss": 0.0252,
+      "step": 21500
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.385408990420044e-07,
+      "loss": 0.2601,
+      "step": 21510
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3839351510685334e-07,
+      "loss": 0.0001,
+      "step": 21520
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.382461311717023e-07,
+      "loss": 0.1414,
+      "step": 21530
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.380987472365512e-07,
+      "loss": 0.1847,
+      "step": 21540
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3795136330140013e-07,
+      "loss": 0.1416,
+      "step": 21550
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3780397936624909e-07,
+      "loss": 0.0002,
+      "step": 21560
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.37656595431098e-07,
+      "loss": 0.4733,
+      "step": 21570
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3750921149594693e-07,
+      "loss": 0.3045,
+      "step": 21580
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3736182756079586e-07,
+      "loss": 0.0004,
+      "step": 21590
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.372144436256448e-07,
+      "loss": 0.1166,
+      "step": 21600
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3706705969049373e-07,
+      "loss": 0.0005,
+      "step": 21610
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3691967575534266e-07,
+      "loss": 0.1504,
+      "step": 21620
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.367722918201916e-07,
+      "loss": 0.0002,
+      "step": 21630
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3662490788504053e-07,
+      "loss": 0.0002,
+      "step": 21640
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3647752394988945e-07,
+      "loss": 0.0001,
+      "step": 21650
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3633014001473838e-07,
+      "loss": 0.1838,
+      "step": 21660
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3618275607958733e-07,
+      "loss": 0.0007,
+      "step": 21670
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3603537214443625e-07,
+      "loss": 0.1514,
+      "step": 21680
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3588798820928518e-07,
+      "loss": 0.0306,
+      "step": 21690
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3574060427413413e-07,
+      "loss": 0.0009,
+      "step": 21700
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3559322033898305e-07,
+      "loss": 0.0002,
+      "step": 21710
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3544583640383197e-07,
+      "loss": 0.0396,
+      "step": 21720
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.352984524686809e-07,
+      "loss": 0.1717,
+      "step": 21730
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3515106853352985e-07,
+      "loss": 0.0988,
+      "step": 21740
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3500368459837877e-07,
+      "loss": 0.181,
+      "step": 21750
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.348563006632277e-07,
+      "loss": 0.0001,
+      "step": 21760
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3470891672807665e-07,
+      "loss": 0.0001,
+      "step": 21770
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3456153279292557e-07,
+      "loss": 0.0001,
+      "step": 21780
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.344141488577745e-07,
+      "loss": 0.0001,
+      "step": 21790
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3426676492262342e-07,
+      "loss": 0.1892,
+      "step": 21800
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.3411938098747237e-07,
+      "loss": 0.0002,
+      "step": 21810
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.339719970523213e-07,
+      "loss": 0.1359,
+      "step": 21820
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.3382461311717022e-07,
+      "loss": 0.0002,
+      "step": 21830
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.3367722918201914e-07,
+      "loss": 0.0016,
+      "step": 21840
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.335298452468681e-07,
+      "loss": 0.0161,
+      "step": 21850
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.3338246131171701e-07,
+      "loss": 0.0001,
+      "step": 21860
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.3323507737656594e-07,
+      "loss": 0.2609,
+      "step": 21870
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.330876934414149e-07,
+      "loss": 0.0001,
+      "step": 21880
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.329403095062638e-07,
+      "loss": 0.2665,
+      "step": 21890
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.3279292557111274e-07,
+      "loss": 0.2545,
+      "step": 21900
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.3264554163596166e-07,
+      "loss": 0.0733,
+      "step": 21910
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.324981577008106e-07,
+      "loss": 0.4083,
+      "step": 21920
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.3235077376565954e-07,
+      "loss": 0.0001,
+      "step": 21930
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.3220338983050846e-07,
+      "loss": 0.0001,
+      "step": 21940
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.320560058953574e-07,
+      "loss": 0.1749,
+      "step": 21950
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.3190862196020633e-07,
+      "loss": 0.0002,
+      "step": 21960
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.3176123802505526e-07,
+      "loss": 0.1485,
+      "step": 21970
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.3161385408990418e-07,
+      "loss": 0.0961,
+      "step": 21980
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.3146647015475313e-07,
+      "loss": 0.4043,
+      "step": 21990
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.3131908621960206e-07,
+      "loss": 0.1573,
+      "step": 22000
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.3117170228445098e-07,
+      "loss": 0.2144,
+      "step": 22010
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.3102431834929993e-07,
+      "loss": 0.0001,
+      "step": 22020
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.3087693441414885e-07,
+      "loss": 0.0002,
+      "step": 22030
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.3072955047899778e-07,
+      "loss": 0.0001,
+      "step": 22040
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.305821665438467e-07,
+      "loss": 0.2032,
+      "step": 22050
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.3043478260869565e-07,
+      "loss": 0.141,
+      "step": 22060
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.3028739867354458e-07,
+      "loss": 0.0001,
+      "step": 22070
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.301400147383935e-07,
+      "loss": 0.0002,
+      "step": 22080
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.2999263080324245e-07,
+      "loss": 0.1833,
+      "step": 22090
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.2984524686809137e-07,
+      "loss": 0.0001,
+      "step": 22100
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.296978629329403e-07,
+      "loss": 0.0066,
+      "step": 22110
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.2955047899778922e-07,
+      "loss": 0.1436,
+      "step": 22120
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.2940309506263817e-07,
+      "loss": 0.0001,
+      "step": 22130
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.292557111274871e-07,
+      "loss": 0.1906,
+      "step": 22140
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.2910832719233602e-07,
+      "loss": 0.0002,
+      "step": 22150
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.2896094325718494e-07,
+      "loss": 0.0208,
+      "step": 22160
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.288135593220339e-07,
+      "loss": 0.2097,
+      "step": 22170
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.2866617538688282e-07,
+      "loss": 0.0001,
+      "step": 22180
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.2851879145173174e-07,
+      "loss": 0.2777,
+      "step": 22190
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.283714075165807e-07,
+      "loss": 0.0001,
+      "step": 22200
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.2822402358142962e-07,
+      "loss": 0.0002,
+      "step": 22210
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.2807663964627854e-07,
+      "loss": 0.1516,
+      "step": 22220
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.2792925571112746e-07,
+      "loss": 0.1568,
+      "step": 22230
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.2778187177597642e-07,
+      "loss": 0.0044,
+      "step": 22240
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.2763448784082534e-07,
+      "loss": 0.2055,
+      "step": 22250
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.2748710390567426e-07,
+      "loss": 0.2571,
+      "step": 22260
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.2733971997052321e-07,
+      "loss": 0.0002,
+      "step": 22270
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.2719233603537214e-07,
+      "loss": 0.0004,
+      "step": 22280
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.2704495210022106e-07,
+      "loss": 0.0001,
+      "step": 22290
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.2689756816506999e-07,
+      "loss": 0.0002,
+      "step": 22300
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.2675018422991894e-07,
+      "loss": 0.1623,
+      "step": 22310
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.2660280029476786e-07,
+      "loss": 0.0002,
+      "step": 22320
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.2645541635961678e-07,
+      "loss": 0.0002,
+      "step": 22330
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.2630803242446573e-07,
+      "loss": 0.1742,
+      "step": 22340
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.2616064848931466e-07,
+      "loss": 0.2225,
+      "step": 22350
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.2601326455416358e-07,
+      "loss": 0.0001,
+      "step": 22360
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.258658806190125e-07,
+      "loss": 0.0007,
+      "step": 22370
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.2571849668386146e-07,
+      "loss": 0.1422,
+      "step": 22380
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.2557111274871038e-07,
+      "loss": 0.0473,
+      "step": 22390
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.254237288135593e-07,
+      "loss": 0.0008,
+      "step": 22400
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.2527634487840825e-07,
+      "loss": 0.0001,
+      "step": 22410
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.2512896094325718e-07,
+      "loss": 0.0002,
+      "step": 22420
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.249815770081061e-07,
+      "loss": 0.143,
+      "step": 22430
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.2483419307295503e-07,
+      "loss": 0.0831,
+      "step": 22440
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.2468680913780398e-07,
+      "loss": 0.0569,
+      "step": 22450
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.245394252026529e-07,
+      "loss": 0.0003,
+      "step": 22460
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.2439204126750182e-07,
+      "loss": 0.0002,
+      "step": 22470
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.2424465733235077e-07,
+      "loss": 0.0002,
+      "step": 22480
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.240972733971997e-07,
+      "loss": 0.0386,
+      "step": 22490
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.2394988946204862e-07,
+      "loss": 0.0001,
+      "step": 22500
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.2380250552689755e-07,
+      "loss": 0.174,
+      "step": 22510
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.236551215917465e-07,
+      "loss": 0.07,
+      "step": 22520
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.2350773765659542e-07,
+      "loss": 0.1306,
+      "step": 22530
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.2336035372144434e-07,
+      "loss": 0.0001,
+      "step": 22540
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.2321296978629327e-07,
+      "loss": 0.3445,
+      "step": 22550
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.2306558585114222e-07,
+      "loss": 0.2061,
+      "step": 22560
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.2291820191599114e-07,
+      "loss": 0.2805,
+      "step": 22570
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.2277081798084007e-07,
+      "loss": 0.0001,
+      "step": 22580
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.2262343404568902e-07,
+      "loss": 0.4005,
+      "step": 22590
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.2247605011053794e-07,
+      "loss": 0.0004,
+      "step": 22600
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.2232866617538686e-07,
+      "loss": 0.0004,
+      "step": 22610
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.221812822402358e-07,
+      "loss": 0.201,
+      "step": 22620
+    },
+    {
+      "epoch": 8.07,
+      "eval_accuracy": 0.8906311250490004,
+      "eval_loss": 0.7732029557228088,
+      "eval_runtime": 610.9589,
+      "eval_samples_per_second": 4.175,
+      "eval_steps_per_second": 1.044,
+      "step": 22626
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.2203389830508474e-07,
+      "loss": 0.1752,
+      "step": 22630
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.2188651436993366e-07,
+      "loss": 0.1996,
+      "step": 22640
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.217391304347826e-07,
+      "loss": 0.0012,
+      "step": 22650
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.2159174649963154e-07,
+      "loss": 0.1402,
+      "step": 22660
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.2144436256448046e-07,
+      "loss": 0.0004,
+      "step": 22670
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.2129697862932939e-07,
+      "loss": 0.238,
+      "step": 22680
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.211495946941783e-07,
+      "loss": 0.1826,
+      "step": 22690
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.2100221075902726e-07,
+      "loss": 0.1597,
+      "step": 22700
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.2085482682387618e-07,
+      "loss": 0.0001,
+      "step": 22710
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.207074428887251e-07,
+      "loss": 0.0001,
+      "step": 22720
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.2056005895357406e-07,
+      "loss": 0.2498,
+      "step": 22730
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.2041267501842298e-07,
+      "loss": 0.1738,
+      "step": 22740
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.202652910832719e-07,
+      "loss": 0.2094,
+      "step": 22750
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.2011790714812083e-07,
+      "loss": 0.0003,
+      "step": 22760
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.1997052321296978e-07,
+      "loss": 0.0014,
+      "step": 22770
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.198231392778187e-07,
+      "loss": 0.0061,
+      "step": 22780
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.1967575534266763e-07,
+      "loss": 0.0016,
+      "step": 22790
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.1952837140751658e-07,
+      "loss": 0.0012,
+      "step": 22800
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.193809874723655e-07,
+      "loss": 0.0031,
+      "step": 22810
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.1923360353721443e-07,
+      "loss": 0.0978,
+      "step": 22820
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.1908621960206335e-07,
+      "loss": 0.1087,
+      "step": 22830
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.189388356669123e-07,
+      "loss": 0.0001,
+      "step": 22840
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.1879145173176122e-07,
+      "loss": 0.1971,
+      "step": 22850
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.1864406779661015e-07,
+      "loss": 0.3327,
+      "step": 22860
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.184966838614591e-07,
+      "loss": 0.2596,
+      "step": 22870
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.1834929992630802e-07,
+      "loss": 0.0795,
+      "step": 22880
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.1820191599115695e-07,
+      "loss": 0.0001,
+      "step": 22890
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.1805453205600587e-07,
+      "loss": 0.0468,
+      "step": 22900
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.1790714812085482e-07,
+      "loss": 0.0001,
+      "step": 22910
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.1775976418570374e-07,
+      "loss": 0.1285,
+      "step": 22920
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.1761238025055267e-07,
+      "loss": 0.0002,
+      "step": 22930
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.174649963154016e-07,
+      "loss": 0.2142,
+      "step": 22940
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.1731761238025054e-07,
+      "loss": 0.1763,
+      "step": 22950
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.1717022844509947e-07,
+      "loss": 0.0001,
+      "step": 22960
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.170228445099484e-07,
+      "loss": 0.0002,
+      "step": 22970
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.1687546057479734e-07,
+      "loss": 0.0001,
+      "step": 22980
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.1672807663964627e-07,
+      "loss": 0.0014,
+      "step": 22990
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.165806927044952e-07,
+      "loss": 0.1337,
+      "step": 23000
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.164333087693441e-07,
+      "loss": 0.2179,
+      "step": 23010
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.1628592483419306e-07,
+      "loss": 0.007,
+      "step": 23020
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.16138540899042e-07,
+      "loss": 0.1758,
+      "step": 23030
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.159911569638909e-07,
+      "loss": 0.0019,
+      "step": 23040
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.1584377302873986e-07,
+      "loss": 0.0021,
+      "step": 23050
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.1569638909358879e-07,
+      "loss": 0.0001,
+      "step": 23060
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.155490051584377e-07,
+      "loss": 0.0001,
+      "step": 23070
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.1540162122328663e-07,
+      "loss": 0.0001,
+      "step": 23080
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.1525423728813558e-07,
+      "loss": 0.0002,
+      "step": 23090
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.151068533529845e-07,
+      "loss": 0.0003,
+      "step": 23100
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.1495946941783343e-07,
+      "loss": 0.0726,
+      "step": 23110
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.1481208548268238e-07,
+      "loss": 0.0001,
+      "step": 23120
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.146647015475313e-07,
+      "loss": 0.0448,
+      "step": 23130
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.1451731761238023e-07,
+      "loss": 0.0006,
+      "step": 23140
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.1436993367722915e-07,
+      "loss": 0.0007,
+      "step": 23150
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.142225497420781e-07,
+      "loss": 0.0003,
+      "step": 23160
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.1407516580692703e-07,
+      "loss": 0.0001,
+      "step": 23170
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.1392778187177595e-07,
+      "loss": 0.0001,
+      "step": 23180
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.137803979366249e-07,
+      "loss": 0.1907,
+      "step": 23190
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.1363301400147383e-07,
+      "loss": 0.0001,
+      "step": 23200
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.1348563006632275e-07,
+      "loss": 0.0004,
+      "step": 23210
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.1333824613117167e-07,
+      "loss": 0.0001,
+      "step": 23220
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.1319086219602062e-07,
+      "loss": 0.3109,
+      "step": 23230
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.1304347826086955e-07,
+      "loss": 0.0003,
+      "step": 23240
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.1289609432571847e-07,
+      "loss": 0.0012,
+      "step": 23250
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.127487103905674e-07,
+      "loss": 0.0002,
+      "step": 23260
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.1260132645541635e-07,
+      "loss": 0.0002,
+      "step": 23270
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.1245394252026527e-07,
+      "loss": 0.193,
+      "step": 23280
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.123065585851142e-07,
+      "loss": 0.0007,
+      "step": 23290
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.1215917464996315e-07,
+      "loss": 0.0011,
+      "step": 23300
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.1201179071481207e-07,
+      "loss": 0.0004,
+      "step": 23310
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.11864406779661e-07,
+      "loss": 0.0003,
+      "step": 23320
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.1171702284450992e-07,
+      "loss": 0.1794,
+      "step": 23330
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.1156963890935887e-07,
+      "loss": 0.3773,
+      "step": 23340
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.114222549742078e-07,
+      "loss": 0.0001,
+      "step": 23350
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.1127487103905672e-07,
+      "loss": 0.0001,
+      "step": 23360
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.1112748710390567e-07,
+      "loss": 0.2504,
+      "step": 23370
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.109801031687546e-07,
+      "loss": 0.0001,
+      "step": 23380
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.1083271923360351e-07,
+      "loss": 0.001,
+      "step": 23390
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.1068533529845244e-07,
+      "loss": 0.0793,
+      "step": 23400
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.105379513633014e-07,
+      "loss": 0.0001,
+      "step": 23410
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.103905674281503e-07,
+      "loss": 0.166,
+      "step": 23420
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.1024318349299924e-07,
+      "loss": 0.2089,
+      "step": 23430
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.1009579955784819e-07,
+      "loss": 0.0258,
+      "step": 23440
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.099484156226971e-07,
+      "loss": 0.2957,
+      "step": 23450
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.0980103168754603e-07,
+      "loss": 0.2684,
+      "step": 23460
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.0965364775239496e-07,
+      "loss": 0.3242,
+      "step": 23470
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.095062638172439e-07,
+      "loss": 0.1719,
+      "step": 23480
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.0935887988209283e-07,
+      "loss": 0.1583,
+      "step": 23490
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.0921149594694176e-07,
+      "loss": 0.196,
+      "step": 23500
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.090641120117907e-07,
+      "loss": 0.0002,
+      "step": 23510
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.0891672807663963e-07,
+      "loss": 0.0001,
+      "step": 23520
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.0876934414148855e-07,
+      "loss": 0.1794,
+      "step": 23530
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.0862196020633748e-07,
+      "loss": 0.1853,
+      "step": 23540
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.0847457627118643e-07,
+      "loss": 0.0001,
+      "step": 23550
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.0832719233603535e-07,
+      "loss": 0.0826,
+      "step": 23560
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.0817980840088428e-07,
+      "loss": 0.0001,
+      "step": 23570
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.0803242446573323e-07,
+      "loss": 0.0002,
+      "step": 23580
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.0788504053058215e-07,
+      "loss": 0.1578,
+      "step": 23590
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.0773765659543107e-07,
+      "loss": 0.1835,
+      "step": 23600
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.0759027266028e-07,
+      "loss": 0.1605,
+      "step": 23610
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.0744288872512895e-07,
+      "loss": 0.0002,
+      "step": 23620
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.0729550478997787e-07,
+      "loss": 0.0007,
+      "step": 23630
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.071481208548268e-07,
+      "loss": 0.2176,
+      "step": 23640
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.0700073691967572e-07,
+      "loss": 0.1332,
+      "step": 23650
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.0685335298452467e-07,
+      "loss": 0.0001,
+      "step": 23660
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.067059690493736e-07,
+      "loss": 0.2023,
+      "step": 23670
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.0655858511422252e-07,
+      "loss": 0.0022,
+      "step": 23680
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.0641120117907147e-07,
+      "loss": 0.0003,
+      "step": 23690
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.062638172439204e-07,
+      "loss": 0.0003,
+      "step": 23700
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.0611643330876932e-07,
+      "loss": 0.1996,
+      "step": 23710
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.0596904937361824e-07,
+      "loss": 0.0002,
+      "step": 23720
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.058216654384672e-07,
+      "loss": 0.1224,
+      "step": 23730
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.0567428150331612e-07,
+      "loss": 0.1019,
+      "step": 23740
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.0552689756816504e-07,
+      "loss": 0.0002,
+      "step": 23750
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.05379513633014e-07,
+      "loss": 0.0893,
+      "step": 23760
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.0523212969786291e-07,
+      "loss": 0.134,
+      "step": 23770
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.0508474576271184e-07,
+      "loss": 0.0004,
+      "step": 23780
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.0493736182756076e-07,
+      "loss": 0.0033,
+      "step": 23790
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.047899778924097e-07,
+      "loss": 0.0009,
+      "step": 23800
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.0464259395725864e-07,
+      "loss": 0.1007,
+      "step": 23810
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.0449521002210759e-07,
+      "loss": 0.0874,
+      "step": 23820
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.0434782608695654e-07,
+      "loss": 0.0003,
+      "step": 23830
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.0420044215180546e-07,
+      "loss": 0.1594,
+      "step": 23840
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.0405305821665438e-07,
+      "loss": 0.1205,
+      "step": 23850
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.039056742815033e-07,
+      "loss": 0.0003,
+      "step": 23860
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.0375829034635226e-07,
+      "loss": 0.0092,
+      "step": 23870
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.0361090641120118e-07,
+      "loss": 0.1539,
+      "step": 23880
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.034635224760501e-07,
+      "loss": 0.0001,
+      "step": 23890
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.0331613854089906e-07,
+      "loss": 0.0914,
+      "step": 23900
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.0316875460574798e-07,
+      "loss": 0.0003,
+      "step": 23910
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.030213706705969e-07,
+      "loss": 0.0565,
+      "step": 23920
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.0287398673544583e-07,
+      "loss": 0.0004,
+      "step": 23930
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.0272660280029478e-07,
+      "loss": 0.0001,
+      "step": 23940
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.025792188651437e-07,
+      "loss": 0.0003,
+      "step": 23950
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.0243183492999263e-07,
+      "loss": 0.0002,
+      "step": 23960
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.0228445099484158e-07,
+      "loss": 0.0001,
+      "step": 23970
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.021370670596905e-07,
+      "loss": 0.0001,
+      "step": 23980
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.0198968312453943e-07,
+      "loss": 0.1634,
+      "step": 23990
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.0184229918938835e-07,
+      "loss": 0.0003,
+      "step": 24000
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.016949152542373e-07,
+      "loss": 0.0002,
+      "step": 24010
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.0154753131908622e-07,
+      "loss": 0.0001,
+      "step": 24020
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.0140014738393515e-07,
+      "loss": 0.0001,
+      "step": 24030
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.012527634487841e-07,
+      "loss": 0.0003,
+      "step": 24040
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.0110537951363302e-07,
+      "loss": 0.0001,
+      "step": 24050
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.0095799557848195e-07,
+      "loss": 0.3671,
+      "step": 24060
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.0081061164333087e-07,
+      "loss": 0.0867,
+      "step": 24070
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.0066322770817982e-07,
+      "loss": 0.0006,
+      "step": 24080
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.0051584377302874e-07,
+      "loss": 0.0001,
+      "step": 24090
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.0036845983787767e-07,
+      "loss": 0.3,
+      "step": 24100
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.002210759027266e-07,
+      "loss": 0.0006,
+      "step": 24110
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.0007369196757554e-07,
+      "loss": 0.013,
+      "step": 24120
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.9992630803242447e-07,
+      "loss": 0.0003,
+      "step": 24130
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.997789240972734e-07,
+      "loss": 0.0264,
+      "step": 24140
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.9963154016212234e-07,
+      "loss": 0.0001,
+      "step": 24150
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.9948415622697126e-07,
+      "loss": 0.25,
+      "step": 24160
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.993367722918202e-07,
+      "loss": 0.0003,
+      "step": 24170
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.991893883566691e-07,
+      "loss": 0.0011,
+      "step": 24180
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.9904200442151806e-07,
+      "loss": 0.1779,
+      "step": 24190
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.9889462048636699e-07,
+      "loss": 0.0001,
+      "step": 24200
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.987472365512159e-07,
+      "loss": 0.0001,
+      "step": 24210
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.9859985261606486e-07,
+      "loss": 0.0003,
+      "step": 24220
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.9845246868091378e-07,
+      "loss": 0.0862,
+      "step": 24230
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.983050847457627e-07,
+      "loss": 0.1784,
+      "step": 24240
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.9815770081061163e-07,
+      "loss": 0.2115,
+      "step": 24250
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.9801031687546058e-07,
+      "loss": 0.002,
+      "step": 24260
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.978629329403095e-07,
+      "loss": 0.577,
+      "step": 24270
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.9771554900515843e-07,
+      "loss": 0.3918,
+      "step": 24280
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.9756816507000738e-07,
+      "loss": 0.1672,
+      "step": 24290
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.974207811348563e-07,
+      "loss": 0.0001,
+      "step": 24300
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.9727339719970523e-07,
+      "loss": 0.0001,
+      "step": 24310
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.9712601326455415e-07,
+      "loss": 0.0,
+      "step": 24320
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.969786293294031e-07,
+      "loss": 0.0001,
+      "step": 24330
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.9683124539425203e-07,
+      "loss": 0.204,
+      "step": 24340
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.9668386145910095e-07,
+      "loss": 0.0001,
+      "step": 24350
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.965364775239499e-07,
+      "loss": 0.1873,
+      "step": 24360
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.9638909358879883e-07,
+      "loss": 0.0001,
+      "step": 24370
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.9624170965364775e-07,
+      "loss": 0.0001,
+      "step": 24380
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.9609432571849667e-07,
+      "loss": 0.0821,
+      "step": 24390
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.9594694178334562e-07,
+      "loss": 0.3263,
+      "step": 24400
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.9579955784819455e-07,
+      "loss": 0.0179,
+      "step": 24410
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.9565217391304347e-07,
+      "loss": 0.1337,
+      "step": 24420
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.9550478997789242e-07,
+      "loss": 0.0615,
+      "step": 24430
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.9535740604274135e-07,
+      "loss": 0.0003,
+      "step": 24440
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.9521002210759027e-07,
+      "loss": 0.1313,
+      "step": 24450
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.950626381724392e-07,
+      "loss": 0.0004,
+      "step": 24460
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.9491525423728814e-07,
+      "loss": 0.1347,
+      "step": 24470
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.9476787030213707e-07,
+      "loss": 0.1521,
+      "step": 24480
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.94620486366986e-07,
+      "loss": 0.0002,
+      "step": 24490
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.9447310243183492e-07,
+      "loss": 0.237,
+      "step": 24500
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.9432571849668387e-07,
+      "loss": 0.0002,
+      "step": 24510
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.941783345615328e-07,
+      "loss": 0.0001,
+      "step": 24520
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.9403095062638171e-07,
+      "loss": 0.0024,
+      "step": 24530
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.9388356669123066e-07,
+      "loss": 0.0046,
+      "step": 24540
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.937361827560796e-07,
+      "loss": 0.1592,
+      "step": 24550
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.935887988209285e-07,
+      "loss": 0.0001,
+      "step": 24560
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.9344141488577744e-07,
+      "loss": 0.1789,
+      "step": 24570
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.932940309506264e-07,
+      "loss": 0.0289,
+      "step": 24580
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.931466470154753e-07,
+      "loss": 0.2239,
+      "step": 24590
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.9299926308032423e-07,
+      "loss": 0.0006,
+      "step": 24600
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.9285187914517319e-07,
+      "loss": 0.2025,
+      "step": 24610
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.927044952100221e-07,
+      "loss": 0.0003,
+      "step": 24620
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.9255711127487103e-07,
+      "loss": 0.2329,
+      "step": 24630
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.9240972733971996e-07,
+      "loss": 0.0186,
+      "step": 24640
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.922623434045689e-07,
+      "loss": 0.139,
+      "step": 24650
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.9211495946941783e-07,
+      "loss": 0.0001,
+      "step": 24660
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.9196757553426676e-07,
+      "loss": 0.0183,
+      "step": 24670
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.918201915991157e-07,
+      "loss": 0.0002,
+      "step": 24680
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.9167280766396463e-07,
+      "loss": 0.0002,
+      "step": 24690
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.9152542372881355e-07,
+      "loss": 0.5882,
+      "step": 24700
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.9137803979366248e-07,
+      "loss": 0.0002,
+      "step": 24710
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.9123065585851143e-07,
+      "loss": 0.0001,
+      "step": 24720
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.9108327192336035e-07,
+      "loss": 0.0001,
+      "step": 24730
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.9093588798820928e-07,
+      "loss": 0.2162,
+      "step": 24740
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.9078850405305823e-07,
+      "loss": 0.0002,
+      "step": 24750
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.9064112011790715e-07,
+      "loss": 0.0707,
+      "step": 24760
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.9049373618275607e-07,
+      "loss": 0.0014,
+      "step": 24770
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.90346352247605e-07,
+      "loss": 0.0001,
+      "step": 24780
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.9019896831245395e-07,
+      "loss": 0.2092,
+      "step": 24790
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.9005158437730287e-07,
+      "loss": 0.0001,
+      "step": 24800
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.899042004421518e-07,
+      "loss": 0.0001,
+      "step": 24810
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.8975681650700075e-07,
+      "loss": 0.0002,
+      "step": 24820
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.8960943257184967e-07,
+      "loss": 0.0001,
+      "step": 24830
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.894620486366986e-07,
+      "loss": 0.0001,
+      "step": 24840
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.8931466470154752e-07,
+      "loss": 0.0013,
+      "step": 24850
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.8916728076639647e-07,
+      "loss": 0.3904,
+      "step": 24860
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.890198968312454e-07,
+      "loss": 0.0001,
+      "step": 24870
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.8887251289609432e-07,
+      "loss": 0.241,
+      "step": 24880
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.8872512896094324e-07,
+      "loss": 0.0004,
+      "step": 24890
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.885777450257922e-07,
+      "loss": 0.0108,
+      "step": 24900
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.8843036109064111e-07,
+      "loss": 0.001,
+      "step": 24910
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.8828297715549004e-07,
+      "loss": 0.1441,
+      "step": 24920
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.88135593220339e-07,
+      "loss": 0.0002,
+      "step": 24930
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.879882092851879e-07,
+      "loss": 0.0001,
+      "step": 24940
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.8784082535003684e-07,
+      "loss": 0.0007,
+      "step": 24950
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.8769344141488576e-07,
+      "loss": 0.0708,
+      "step": 24960
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.875460574797347e-07,
+      "loss": 0.3231,
+      "step": 24970
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.8739867354458363e-07,
+      "loss": 0.0003,
+      "step": 24980
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.8725128960943256e-07,
+      "loss": 0.1111,
+      "step": 24990
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.871039056742815e-07,
+      "loss": 0.0011,
+      "step": 25000
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.8695652173913043e-07,
+      "loss": 0.0001,
+      "step": 25010
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.8680913780397936e-07,
+      "loss": 0.0001,
+      "step": 25020
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.8666175386882828e-07,
+      "loss": 0.0002,
+      "step": 25030
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.8651436993367723e-07,
+      "loss": 0.0001,
+      "step": 25040
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.8636698599852616e-07,
+      "loss": 0.0001,
+      "step": 25050
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.8621960206337508e-07,
+      "loss": 0.142,
+      "step": 25060
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.8607221812822403e-07,
+      "loss": 0.1336,
+      "step": 25070
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.8592483419307295e-07,
+      "loss": 0.0021,
+      "step": 25080
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.8577745025792188e-07,
+      "loss": 0.0332,
+      "step": 25090
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.856300663227708e-07,
+      "loss": 0.0001,
+      "step": 25100
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.8548268238761975e-07,
+      "loss": 0.1755,
+      "step": 25110
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.8533529845246868e-07,
+      "loss": 0.2772,
+      "step": 25120
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.851879145173176e-07,
+      "loss": 0.1813,
+      "step": 25130
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.8504053058216655e-07,
+      "loss": 0.0237,
+      "step": 25140
+    },
+    {
+      "epoch": 9.07,
+      "eval_accuracy": 0.8992551940415523,
+      "eval_loss": 0.7024816274642944,
+      "eval_runtime": 593.5204,
+      "eval_samples_per_second": 4.298,
+      "eval_steps_per_second": 1.075,
+      "step": 25140
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.8489314664701547e-07,
+      "loss": 0.1262,
+      "step": 25150
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.847457627118644e-07,
+      "loss": 0.0001,
+      "step": 25160
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.8459837877671332e-07,
+      "loss": 0.0001,
+      "step": 25170
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.8445099484156227e-07,
+      "loss": 0.2314,
+      "step": 25180
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.843036109064112e-07,
+      "loss": 0.0002,
+      "step": 25190
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.8415622697126012e-07,
+      "loss": 0.0001,
+      "step": 25200
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.8400884303610904e-07,
+      "loss": 0.1168,
+      "step": 25210
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.83861459100958e-07,
+      "loss": 0.0001,
+      "step": 25220
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.8371407516580692e-07,
+      "loss": 0.0001,
+      "step": 25230
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.8356669123065584e-07,
+      "loss": 0.2785,
+      "step": 25240
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.834193072955048e-07,
+      "loss": 0.0001,
+      "step": 25250
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.8327192336035372e-07,
+      "loss": 0.0001,
+      "step": 25260
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.8312453942520264e-07,
+      "loss": 0.0003,
+      "step": 25270
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.8297715549005156e-07,
+      "loss": 0.0001,
+      "step": 25280
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.8282977155490051e-07,
+      "loss": 0.0001,
+      "step": 25290
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.8268238761974944e-07,
+      "loss": 0.0001,
+      "step": 25300
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.8253500368459836e-07,
+      "loss": 0.3866,
+      "step": 25310
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.823876197494473e-07,
+      "loss": 0.0747,
+      "step": 25320
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.8224023581429624e-07,
+      "loss": 0.0008,
+      "step": 25330
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.8209285187914516e-07,
+      "loss": 0.0002,
+      "step": 25340
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.8194546794399408e-07,
+      "loss": 0.0117,
+      "step": 25350
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.8179808400884304e-07,
+      "loss": 0.0001,
+      "step": 25360
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.8165070007369196e-07,
+      "loss": 0.0001,
+      "step": 25370
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.8150331613854088e-07,
+      "loss": 0.0138,
+      "step": 25380
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.8135593220338983e-07,
+      "loss": 0.0001,
+      "step": 25390
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.8120854826823876e-07,
+      "loss": 0.0002,
+      "step": 25400
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.8106116433308768e-07,
+      "loss": 0.0003,
+      "step": 25410
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.809137803979366e-07,
+      "loss": 0.0001,
+      "step": 25420
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.8076639646278556e-07,
+      "loss": 0.0005,
+      "step": 25430
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.8061901252763448e-07,
+      "loss": 0.0002,
+      "step": 25440
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.804716285924834e-07,
+      "loss": 0.3455,
+      "step": 25450
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.8032424465733235e-07,
+      "loss": 0.0001,
+      "step": 25460
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.8017686072218128e-07,
+      "loss": 0.1388,
+      "step": 25470
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.800294767870302e-07,
+      "loss": 0.0001,
+      "step": 25480
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.7988209285187913e-07,
+      "loss": 0.0036,
+      "step": 25490
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.7973470891672808e-07,
+      "loss": 0.0019,
+      "step": 25500
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.79587324981577e-07,
+      "loss": 0.0,
+      "step": 25510
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.7943994104642592e-07,
+      "loss": 0.0002,
+      "step": 25520
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.7929255711127487e-07,
+      "loss": 0.092,
+      "step": 25530
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.791451731761238e-07,
+      "loss": 0.2586,
+      "step": 25540
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.7899778924097272e-07,
+      "loss": 0.1309,
+      "step": 25550
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.7885040530582165e-07,
+      "loss": 0.0129,
+      "step": 25560
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.787030213706706e-07,
+      "loss": 0.1993,
+      "step": 25570
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.7855563743551952e-07,
+      "loss": 0.2216,
+      "step": 25580
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.7840825350036844e-07,
+      "loss": 0.0001,
+      "step": 25590
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.7826086956521737e-07,
+      "loss": 0.0001,
+      "step": 25600
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.7811348563006632e-07,
+      "loss": 0.1581,
+      "step": 25610
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.7796610169491524e-07,
+      "loss": 0.0003,
+      "step": 25620
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.7781871775976417e-07,
+      "loss": 0.3325,
+      "step": 25630
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.7767133382461312e-07,
+      "loss": 0.0001,
+      "step": 25640
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.7752394988946204e-07,
+      "loss": 0.1858,
+      "step": 25650
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.7737656595431096e-07,
+      "loss": 0.2797,
+      "step": 25660
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.772291820191599e-07,
+      "loss": 0.0001,
+      "step": 25670
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.7708179808400884e-07,
+      "loss": 0.0975,
+      "step": 25680
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.7693441414885776e-07,
+      "loss": 0.201,
+      "step": 25690
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.767870302137067e-07,
+      "loss": 0.0001,
+      "step": 25700
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.7663964627855564e-07,
+      "loss": 0.0001,
+      "step": 25710
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.7649226234340456e-07,
+      "loss": 0.1864,
+      "step": 25720
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.7634487840825349e-07,
+      "loss": 0.0002,
+      "step": 25730
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.761974944731024e-07,
+      "loss": 0.0001,
+      "step": 25740
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.7605011053795136e-07,
+      "loss": 0.0001,
+      "step": 25750
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.7590272660280028e-07,
+      "loss": 0.0968,
+      "step": 25760
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.757553426676492e-07,
+      "loss": 0.0731,
+      "step": 25770
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.7560795873249816e-07,
+      "loss": 0.0001,
+      "step": 25780
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.7546057479734708e-07,
+      "loss": 0.0004,
+      "step": 25790
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.75313190862196e-07,
+      "loss": 0.0002,
+      "step": 25800
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.7516580692704493e-07,
+      "loss": 0.0001,
+      "step": 25810
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.7501842299189388e-07,
+      "loss": 0.0001,
+      "step": 25820
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.748710390567428e-07,
+      "loss": 0.2036,
+      "step": 25830
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.7472365512159173e-07,
+      "loss": 0.1895,
+      "step": 25840
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.7457627118644068e-07,
+      "loss": 0.1265,
+      "step": 25850
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.744288872512896e-07,
+      "loss": 0.0003,
+      "step": 25860
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.7428150331613853e-07,
+      "loss": 0.0001,
+      "step": 25870
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.7413411938098745e-07,
+      "loss": 0.0001,
+      "step": 25880
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.739867354458364e-07,
+      "loss": 0.1275,
+      "step": 25890
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.7383935151068532e-07,
+      "loss": 0.0802,
+      "step": 25900
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.7369196757553425e-07,
+      "loss": 0.0365,
+      "step": 25910
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.735445836403832e-07,
+      "loss": 0.172,
+      "step": 25920
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.7339719970523212e-07,
+      "loss": 0.0153,
+      "step": 25930
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.7324981577008105e-07,
+      "loss": 0.0011,
+      "step": 25940
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.7310243183492997e-07,
+      "loss": 0.0289,
+      "step": 25950
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.7295504789977892e-07,
+      "loss": 0.1673,
+      "step": 25960
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.7280766396462784e-07,
+      "loss": 0.0001,
+      "step": 25970
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.7266028002947677e-07,
+      "loss": 0.0001,
+      "step": 25980
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.725128960943257e-07,
+      "loss": 0.3903,
+      "step": 25990
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.7236551215917464e-07,
+      "loss": 0.0001,
+      "step": 26000
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.7221812822402357e-07,
+      "loss": 0.0001,
+      "step": 26010
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.720707442888725e-07,
+      "loss": 0.0001,
+      "step": 26020
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.7192336035372144e-07,
+      "loss": 0.0001,
+      "step": 26030
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.7177597641857037e-07,
+      "loss": 0.141,
+      "step": 26040
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.716285924834193e-07,
+      "loss": 0.0001,
+      "step": 26050
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.714812085482682e-07,
+      "loss": 0.004,
+      "step": 26060
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.7133382461311716e-07,
+      "loss": 0.085,
+      "step": 26070
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.711864406779661e-07,
+      "loss": 0.0454,
+      "step": 26080
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.71039056742815e-07,
+      "loss": 0.0001,
+      "step": 26090
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.7089167280766396e-07,
+      "loss": 0.1274,
+      "step": 26100
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.7074428887251289e-07,
+      "loss": 0.0001,
+      "step": 26110
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.705969049373618e-07,
+      "loss": 0.0001,
+      "step": 26120
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.7044952100221073e-07,
+      "loss": 0.0001,
+      "step": 26130
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.7030213706705968e-07,
+      "loss": 0.1071,
+      "step": 26140
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.701547531319086e-07,
+      "loss": 0.0001,
+      "step": 26150
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.7000736919675753e-07,
+      "loss": 0.1821,
+      "step": 26160
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.6985998526160648e-07,
+      "loss": 0.0002,
+      "step": 26170
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.697126013264554e-07,
+      "loss": 0.0001,
+      "step": 26180
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.6956521739130433e-07,
+      "loss": 0.1674,
+      "step": 26190
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.6941783345615325e-07,
+      "loss": 0.1681,
+      "step": 26200
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.692704495210022e-07,
+      "loss": 0.2801,
+      "step": 26210
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.6912306558585113e-07,
+      "loss": 0.0001,
+      "step": 26220
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.6897568165070005e-07,
+      "loss": 0.0002,
+      "step": 26230
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.68828297715549e-07,
+      "loss": 0.0001,
+      "step": 26240
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.6868091378039793e-07,
+      "loss": 0.1418,
+      "step": 26250
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.6853352984524685e-07,
+      "loss": 0.0067,
+      "step": 26260
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.6838614591009577e-07,
+      "loss": 0.0001,
+      "step": 26270
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.6823876197494472e-07,
+      "loss": 0.0001,
+      "step": 26280
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.6809137803979365e-07,
+      "loss": 0.0001,
+      "step": 26290
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.6794399410464257e-07,
+      "loss": 0.1088,
+      "step": 26300
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.677966101694915e-07,
+      "loss": 0.0004,
+      "step": 26310
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.6764922623434045e-07,
+      "loss": 0.0001,
+      "step": 26320
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.6750184229918937e-07,
+      "loss": 0.0001,
+      "step": 26330
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.673544583640383e-07,
+      "loss": 0.0001,
+      "step": 26340
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.6720707442888724e-07,
+      "loss": 0.0528,
+      "step": 26350
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.6705969049373617e-07,
+      "loss": 0.382,
+      "step": 26360
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.669123065585851e-07,
+      "loss": 0.0001,
+      "step": 26370
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.6676492262343402e-07,
+      "loss": 0.0002,
+      "step": 26380
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.6661753868828297e-07,
+      "loss": 0.0041,
+      "step": 26390
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.664701547531319e-07,
+      "loss": 0.0006,
+      "step": 26400
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.6632277081798081e-07,
+      "loss": 0.0017,
+      "step": 26410
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.6617538688282977e-07,
+      "loss": 0.1006,
+      "step": 26420
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.660280029476787e-07,
+      "loss": 0.0192,
+      "step": 26430
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.658806190125276e-07,
+      "loss": 0.195,
+      "step": 26440
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.6573323507737654e-07,
+      "loss": 0.0001,
+      "step": 26450
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.655858511422255e-07,
+      "loss": 0.206,
+      "step": 26460
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.654384672070744e-07,
+      "loss": 0.0001,
+      "step": 26470
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6529108327192334e-07,
+      "loss": 0.0001,
+      "step": 26480
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6514369933677229e-07,
+      "loss": 0.0001,
+      "step": 26490
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.649963154016212e-07,
+      "loss": 0.1333,
+      "step": 26500
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6484893146647013e-07,
+      "loss": 0.337,
+      "step": 26510
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6470154753131906e-07,
+      "loss": 0.0001,
+      "step": 26520
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.64554163596168e-07,
+      "loss": 0.0001,
+      "step": 26530
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6440677966101693e-07,
+      "loss": 0.0102,
+      "step": 26540
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6425939572586586e-07,
+      "loss": 0.0002,
+      "step": 26550
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.641120117907148e-07,
+      "loss": 0.0001,
+      "step": 26560
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6396462785556373e-07,
+      "loss": 0.0046,
+      "step": 26570
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6381724392041265e-07,
+      "loss": 0.0,
+      "step": 26580
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6366985998526158e-07,
+      "loss": 0.164,
+      "step": 26590
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6352247605011053e-07,
+      "loss": 0.0029,
+      "step": 26600
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6337509211495945e-07,
+      "loss": 0.1702,
+      "step": 26610
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6322770817980838e-07,
+      "loss": 0.0001,
+      "step": 26620
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6308032424465733e-07,
+      "loss": 0.0001,
+      "step": 26630
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6293294030950625e-07,
+      "loss": 0.262,
+      "step": 26640
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6278555637435517e-07,
+      "loss": 0.3396,
+      "step": 26650
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.626381724392041e-07,
+      "loss": 0.146,
+      "step": 26660
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6249078850405305e-07,
+      "loss": 0.0313,
+      "step": 26670
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6234340456890197e-07,
+      "loss": 0.0002,
+      "step": 26680
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.621960206337509e-07,
+      "loss": 0.0001,
+      "step": 26690
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6204863669859982e-07,
+      "loss": 0.0001,
+      "step": 26700
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6190125276344877e-07,
+      "loss": 0.0075,
+      "step": 26710
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.617538688282977e-07,
+      "loss": 0.0001,
+      "step": 26720
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6160648489314662e-07,
+      "loss": 0.0012,
+      "step": 26730
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6145910095799557e-07,
+      "loss": 0.0001,
+      "step": 26740
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.613117170228445e-07,
+      "loss": 0.0001,
+      "step": 26750
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6116433308769342e-07,
+      "loss": 0.0001,
+      "step": 26760
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6101694915254234e-07,
+      "loss": 0.0001,
+      "step": 26770
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.608695652173913e-07,
+      "loss": 0.0672,
+      "step": 26780
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6072218128224022e-07,
+      "loss": 0.0001,
+      "step": 26790
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6057479734708914e-07,
+      "loss": 0.0447,
+      "step": 26800
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.604274134119381e-07,
+      "loss": 0.3076,
+      "step": 26810
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6028002947678701e-07,
+      "loss": 0.0001,
+      "step": 26820
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6013264554163594e-07,
+      "loss": 0.187,
+      "step": 26830
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.5998526160648486e-07,
+      "loss": 0.0001,
+      "step": 26840
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.598378776713338e-07,
+      "loss": 0.0001,
+      "step": 26850
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.5969049373618274e-07,
+      "loss": 0.0002,
+      "step": 26860
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.5954310980103166e-07,
+      "loss": 0.3575,
+      "step": 26870
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.593957258658806e-07,
+      "loss": 0.154,
+      "step": 26880
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.5924834193072953e-07,
+      "loss": 0.0001,
+      "step": 26890
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.5910095799557846e-07,
+      "loss": 0.0001,
+      "step": 26900
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.589535740604274e-07,
+      "loss": 0.0001,
+      "step": 26910
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.5880619012527636e-07,
+      "loss": 0.0003,
+      "step": 26920
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.5865880619012528e-07,
+      "loss": 0.1313,
+      "step": 26930
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.585114222549742e-07,
+      "loss": 0.0142,
+      "step": 26940
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.5836403831982316e-07,
+      "loss": 0.1875,
+      "step": 26950
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.5821665438467208e-07,
+      "loss": 0.1019,
+      "step": 26960
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.58069270449521e-07,
+      "loss": 0.0001,
+      "step": 26970
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.5792188651436993e-07,
+      "loss": 0.0752,
+      "step": 26980
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.5777450257921888e-07,
+      "loss": 0.0001,
+      "step": 26990
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.576271186440678e-07,
+      "loss": 0.0004,
+      "step": 27000
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.5747973470891673e-07,
+      "loss": 0.0001,
+      "step": 27010
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.5733235077376568e-07,
+      "loss": 0.0186,
+      "step": 27020
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.571849668386146e-07,
+      "loss": 0.0001,
+      "step": 27030
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.5703758290346353e-07,
+      "loss": 0.0001,
+      "step": 27040
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.5689019896831245e-07,
+      "loss": 0.2056,
+      "step": 27050
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.567428150331614e-07,
+      "loss": 0.0003,
+      "step": 27060
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.5659543109801032e-07,
+      "loss": 0.1178,
+      "step": 27070
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.5644804716285925e-07,
+      "loss": 0.0086,
+      "step": 27080
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.563006632277082e-07,
+      "loss": 0.0002,
+      "step": 27090
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.5615327929255712e-07,
+      "loss": 0.2904,
+      "step": 27100
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.5600589535740605e-07,
+      "loss": 0.0468,
+      "step": 27110
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.5585851142225497e-07,
+      "loss": 0.1006,
+      "step": 27120
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.5571112748710392e-07,
+      "loss": 0.0011,
+      "step": 27130
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.5556374355195284e-07,
+      "loss": 0.0,
+      "step": 27140
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.5541635961680177e-07,
+      "loss": 0.0001,
+      "step": 27150
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.552689756816507e-07,
+      "loss": 0.0012,
+      "step": 27160
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.5512159174649964e-07,
+      "loss": 0.0001,
+      "step": 27170
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.5497420781134857e-07,
+      "loss": 0.0008,
+      "step": 27180
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.548268238761975e-07,
+      "loss": 0.1923,
+      "step": 27190
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.5467943994104644e-07,
+      "loss": 0.0001,
+      "step": 27200
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.5453205600589536e-07,
+      "loss": 0.0417,
+      "step": 27210
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.543846720707443e-07,
+      "loss": 0.2137,
+      "step": 27220
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.542372881355932e-07,
+      "loss": 0.0001,
+      "step": 27230
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.5408990420044216e-07,
+      "loss": 0.0226,
+      "step": 27240
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.5394252026529109e-07,
+      "loss": 0.0,
+      "step": 27250
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.5379513633014e-07,
+      "loss": 0.0,
+      "step": 27260
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.5364775239498896e-07,
+      "loss": 0.0,
+      "step": 27270
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.5350036845983788e-07,
+      "loss": 0.1954,
+      "step": 27280
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.533529845246868e-07,
+      "loss": 0.0003,
+      "step": 27290
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.5320560058953573e-07,
+      "loss": 0.0367,
+      "step": 27300
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.5305821665438468e-07,
+      "loss": 0.0001,
+      "step": 27310
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.529108327192336e-07,
+      "loss": 0.0001,
+      "step": 27320
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.5276344878408253e-07,
+      "loss": 0.0001,
+      "step": 27330
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.5261606484893148e-07,
+      "loss": 0.0001,
+      "step": 27340
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.524686809137804e-07,
+      "loss": 0.1351,
+      "step": 27350
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.5232129697862933e-07,
+      "loss": 0.0001,
+      "step": 27360
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.5217391304347825e-07,
+      "loss": 0.0023,
+      "step": 27370
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.520265291083272e-07,
+      "loss": 0.3952,
+      "step": 27380
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.5187914517317613e-07,
+      "loss": 0.0001,
+      "step": 27390
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.5173176123802505e-07,
+      "loss": 0.187,
+      "step": 27400
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.51584377302874e-07,
+      "loss": 0.0001,
+      "step": 27410
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.5143699336772293e-07,
+      "loss": 0.0001,
+      "step": 27420
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.5128960943257185e-07,
+      "loss": 0.204,
+      "step": 27430
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.5114222549742077e-07,
+      "loss": 0.0294,
+      "step": 27440
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.5099484156226972e-07,
+      "loss": 0.2052,
+      "step": 27450
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.5084745762711865e-07,
+      "loss": 0.1934,
+      "step": 27460
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.5070007369196757e-07,
+      "loss": 0.0002,
+      "step": 27470
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.5055268975681652e-07,
+      "loss": 0.2682,
+      "step": 27480
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.5040530582166545e-07,
+      "loss": 0.1187,
+      "step": 27490
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.5025792188651437e-07,
+      "loss": 0.0001,
+      "step": 27500
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.501105379513633e-07,
+      "loss": 0.3651,
+      "step": 27510
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.4996315401621224e-07,
+      "loss": 0.0001,
+      "step": 27520
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.4981577008106117e-07,
+      "loss": 0.1746,
+      "step": 27530
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.496683861459101e-07,
+      "loss": 0.0001,
+      "step": 27540
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.4952100221075902e-07,
+      "loss": 0.0001,
+      "step": 27550
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.4937361827560797e-07,
+      "loss": 0.0001,
+      "step": 27560
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.492262343404569e-07,
+      "loss": 0.0001,
+      "step": 27570
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.4907885040530581e-07,
+      "loss": 0.0001,
+      "step": 27580
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.4893146647015476e-07,
+      "loss": 0.0001,
+      "step": 27590
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.487840825350037e-07,
+      "loss": 0.0016,
+      "step": 27600
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.486366985998526e-07,
+      "loss": 0.0001,
+      "step": 27610
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.4848931466470154e-07,
+      "loss": 0.0001,
+      "step": 27620
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.4834193072955049e-07,
+      "loss": 0.0001,
+      "step": 27630
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.481945467943994e-07,
+      "loss": 0.0,
+      "step": 27640
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.4804716285924833e-07,
+      "loss": 0.0003,
+      "step": 27650
+    },
+    {
+      "epoch": 10.07,
+      "eval_accuracy": 0.8957271658173266,
+      "eval_loss": 0.7510266304016113,
+      "eval_runtime": 630.9676,
+      "eval_samples_per_second": 4.043,
+      "eval_steps_per_second": 1.011,
+      "step": 27654
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.4789977892409728e-07,
+      "loss": 0.0001,
+      "step": 27660
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.477523949889462e-07,
+      "loss": 0.0001,
+      "step": 27670
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.4760501105379513e-07,
+      "loss": 0.0,
+      "step": 27680
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.4745762711864406e-07,
+      "loss": 0.0,
+      "step": 27690
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.47310243183493e-07,
+      "loss": 0.0001,
+      "step": 27700
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.4716285924834193e-07,
+      "loss": 0.0,
+      "step": 27710
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.4701547531319085e-07,
+      "loss": 0.3466,
+      "step": 27720
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.468680913780398e-07,
+      "loss": 0.0942,
+      "step": 27730
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.4672070744288873e-07,
+      "loss": 0.0581,
+      "step": 27740
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.4657332350773765e-07,
+      "loss": 0.0001,
+      "step": 27750
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.4642593957258658e-07,
+      "loss": 0.1236,
+      "step": 27760
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.4627855563743553e-07,
+      "loss": 0.0001,
+      "step": 27770
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.4613117170228445e-07,
+      "loss": 0.0433,
+      "step": 27780
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.4598378776713338e-07,
+      "loss": 0.0033,
+      "step": 27790
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.4583640383198233e-07,
+      "loss": 0.0001,
+      "step": 27800
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.4568901989683125e-07,
+      "loss": 0.189,
+      "step": 27810
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.4554163596168017e-07,
+      "loss": 0.0001,
+      "step": 27820
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.453942520265291e-07,
+      "loss": 0.0006,
+      "step": 27830
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.4524686809137805e-07,
+      "loss": 0.0009,
+      "step": 27840
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.4509948415622697e-07,
+      "loss": 0.0006,
+      "step": 27850
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.449521002210759e-07,
+      "loss": 0.0021,
+      "step": 27860
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.4480471628592482e-07,
+      "loss": 0.0311,
+      "step": 27870
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.4465733235077377e-07,
+      "loss": 0.0,
+      "step": 27880
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.445099484156227e-07,
+      "loss": 0.0001,
+      "step": 27890
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.4436256448047162e-07,
+      "loss": 0.0002,
+      "step": 27900
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.4421518054532057e-07,
+      "loss": 0.1705,
+      "step": 27910
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.440677966101695e-07,
+      "loss": 0.0004,
+      "step": 27920
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.4392041267501842e-07,
+      "loss": 0.2483,
+      "step": 27930
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.4377302873986734e-07,
+      "loss": 0.0001,
+      "step": 27940
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.436256448047163e-07,
+      "loss": 0.1101,
+      "step": 27950
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.4347826086956521e-07,
+      "loss": 0.0001,
+      "step": 27960
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.4333087693441414e-07,
+      "loss": 0.0001,
+      "step": 27970
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.431834929992631e-07,
+      "loss": 0.0001,
+      "step": 27980
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.43036109064112e-07,
+      "loss": 0.3426,
+      "step": 27990
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.4288872512896094e-07,
+      "loss": 0.0,
+      "step": 28000
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.4274134119380986e-07,
+      "loss": 0.0,
+      "step": 28010
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.425939572586588e-07,
+      "loss": 0.0001,
+      "step": 28020
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.4244657332350773e-07,
+      "loss": 0.0383,
+      "step": 28030
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.4229918938835666e-07,
+      "loss": 0.0007,
+      "step": 28040
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.421518054532056e-07,
+      "loss": 0.0001,
+      "step": 28050
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.4200442151805453e-07,
+      "loss": 0.0001,
+      "step": 28060
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.4185703758290346e-07,
+      "loss": 0.0001,
+      "step": 28070
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.4170965364775238e-07,
+      "loss": 0.1964,
+      "step": 28080
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.4156226971260133e-07,
+      "loss": 0.0,
+      "step": 28090
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.4141488577745026e-07,
+      "loss": 0.2052,
+      "step": 28100
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.4126750184229918e-07,
+      "loss": 0.0001,
+      "step": 28110
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.4112011790714813e-07,
+      "loss": 0.0002,
+      "step": 28120
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.4097273397199705e-07,
+      "loss": 0.0,
+      "step": 28130
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.4082535003684598e-07,
+      "loss": 0.0001,
+      "step": 28140
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.406779661016949e-07,
+      "loss": 0.2055,
+      "step": 28150
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.4053058216654385e-07,
+      "loss": 0.0008,
+      "step": 28160
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.4038319823139278e-07,
+      "loss": 0.0001,
+      "step": 28170
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.402358142962417e-07,
+      "loss": 0.0002,
+      "step": 28180
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.4008843036109065e-07,
+      "loss": 0.0001,
+      "step": 28190
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.3994104642593957e-07,
+      "loss": 0.0001,
+      "step": 28200
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.397936624907885e-07,
+      "loss": 0.0001,
+      "step": 28210
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3964627855563742e-07,
+      "loss": 0.0023,
+      "step": 28220
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3949889462048637e-07,
+      "loss": 0.0001,
+      "step": 28230
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.393515106853353e-07,
+      "loss": 0.0238,
+      "step": 28240
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3920412675018422e-07,
+      "loss": 0.0028,
+      "step": 28250
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3905674281503314e-07,
+      "loss": 0.0001,
+      "step": 28260
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.389093588798821e-07,
+      "loss": 0.0001,
+      "step": 28270
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3876197494473102e-07,
+      "loss": 0.1674,
+      "step": 28280
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3861459100957994e-07,
+      "loss": 0.0001,
+      "step": 28290
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.384672070744289e-07,
+      "loss": 0.0001,
+      "step": 28300
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3831982313927782e-07,
+      "loss": 0.0002,
+      "step": 28310
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3817243920412674e-07,
+      "loss": 0.0001,
+      "step": 28320
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3802505526897566e-07,
+      "loss": 0.1672,
+      "step": 28330
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3787767133382461e-07,
+      "loss": 0.0,
+      "step": 28340
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3773028739867354e-07,
+      "loss": 0.0001,
+      "step": 28350
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3758290346352246e-07,
+      "loss": 0.1708,
+      "step": 28360
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.374355195283714e-07,
+      "loss": 0.0002,
+      "step": 28370
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3728813559322034e-07,
+      "loss": 0.0,
+      "step": 28380
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3714075165806926e-07,
+      "loss": 0.0483,
+      "step": 28390
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3699336772291818e-07,
+      "loss": 0.1816,
+      "step": 28400
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3684598378776714e-07,
+      "loss": 0.1704,
+      "step": 28410
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3669859985261606e-07,
+      "loss": 0.0002,
+      "step": 28420
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3655121591746498e-07,
+      "loss": 0.0001,
+      "step": 28430
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3640383198231393e-07,
+      "loss": 0.0,
+      "step": 28440
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3625644804716286e-07,
+      "loss": 0.1538,
+      "step": 28450
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3610906411201178e-07,
+      "loss": 0.0001,
+      "step": 28460
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.359616801768607e-07,
+      "loss": 0.0002,
+      "step": 28470
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3581429624170966e-07,
+      "loss": 0.0,
+      "step": 28480
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3566691230655858e-07,
+      "loss": 0.0008,
+      "step": 28490
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.355195283714075e-07,
+      "loss": 0.0001,
+      "step": 28500
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3537214443625645e-07,
+      "loss": 0.0001,
+      "step": 28510
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3522476050110538e-07,
+      "loss": 0.0,
+      "step": 28520
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.350773765659543e-07,
+      "loss": 0.0,
+      "step": 28530
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3492999263080323e-07,
+      "loss": 0.0,
+      "step": 28540
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3478260869565218e-07,
+      "loss": 0.0972,
+      "step": 28550
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.346352247605011e-07,
+      "loss": 0.0,
+      "step": 28560
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3448784082535002e-07,
+      "loss": 0.0001,
+      "step": 28570
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3434045689019897e-07,
+      "loss": 0.0067,
+      "step": 28580
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.341930729550479e-07,
+      "loss": 0.0001,
+      "step": 28590
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3404568901989682e-07,
+      "loss": 0.0437,
+      "step": 28600
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3389830508474575e-07,
+      "loss": 0.1221,
+      "step": 28610
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.337509211495947e-07,
+      "loss": 0.0001,
+      "step": 28620
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3360353721444362e-07,
+      "loss": 0.0001,
+      "step": 28630
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3345615327929254e-07,
+      "loss": 0.0033,
+      "step": 28640
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3330876934414147e-07,
+      "loss": 0.0001,
+      "step": 28650
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3316138540899042e-07,
+      "loss": 0.0002,
+      "step": 28660
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3301400147383934e-07,
+      "loss": 0.0036,
+      "step": 28670
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3286661753868827e-07,
+      "loss": 0.0001,
+      "step": 28680
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3271923360353722e-07,
+      "loss": 0.0001,
+      "step": 28690
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3257184966838614e-07,
+      "loss": 0.1774,
+      "step": 28700
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3242446573323506e-07,
+      "loss": 0.0001,
+      "step": 28710
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.32277081798084e-07,
+      "loss": 0.0083,
+      "step": 28720
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3212969786293294e-07,
+      "loss": 0.0001,
+      "step": 28730
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3198231392778186e-07,
+      "loss": 0.0638,
+      "step": 28740
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3183492999263079e-07,
+      "loss": 0.0001,
+      "step": 28750
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3168754605747974e-07,
+      "loss": 0.0116,
+      "step": 28760
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3154016212232866e-07,
+      "loss": 0.1142,
+      "step": 28770
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3139277818717758e-07,
+      "loss": 0.0001,
+      "step": 28780
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.312453942520265e-07,
+      "loss": 0.0001,
+      "step": 28790
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3109801031687546e-07,
+      "loss": 0.0083,
+      "step": 28800
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3095062638172438e-07,
+      "loss": 0.0001,
+      "step": 28810
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.308032424465733e-07,
+      "loss": 0.1346,
+      "step": 28820
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3065585851142226e-07,
+      "loss": 0.0,
+      "step": 28830
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3050847457627118e-07,
+      "loss": 0.0087,
+      "step": 28840
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.303610906411201e-07,
+      "loss": 0.0001,
+      "step": 28850
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3021370670596903e-07,
+      "loss": 0.0,
+      "step": 28860
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3006632277081798e-07,
+      "loss": 0.0,
+      "step": 28870
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.299189388356669e-07,
+      "loss": 0.2044,
+      "step": 28880
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.2977155490051583e-07,
+      "loss": 0.113,
+      "step": 28890
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.2962417096536478e-07,
+      "loss": 0.2132,
+      "step": 28900
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.294767870302137e-07,
+      "loss": 0.0349,
+      "step": 28910
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.2932940309506263e-07,
+      "loss": 0.0,
+      "step": 28920
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.2918201915991155e-07,
+      "loss": 0.1692,
+      "step": 28930
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.290346352247605e-07,
+      "loss": 0.0914,
+      "step": 28940
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.2888725128960942e-07,
+      "loss": 0.1393,
+      "step": 28950
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.2873986735445835e-07,
+      "loss": 0.0002,
+      "step": 28960
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.2859248341930727e-07,
+      "loss": 0.0001,
+      "step": 28970
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.2844509948415622e-07,
+      "loss": 0.0103,
+      "step": 28980
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.2829771554900515e-07,
+      "loss": 0.0,
+      "step": 28990
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.2815033161385407e-07,
+      "loss": 0.1615,
+      "step": 29000
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.2800294767870302e-07,
+      "loss": 0.0163,
+      "step": 29010
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.2785556374355194e-07,
+      "loss": 0.0,
+      "step": 29020
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.2770817980840087e-07,
+      "loss": 0.0001,
+      "step": 29030
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.275607958732498e-07,
+      "loss": 0.0001,
+      "step": 29040
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.2741341193809874e-07,
+      "loss": 0.0001,
+      "step": 29050
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.2726602800294767e-07,
+      "loss": 0.0,
+      "step": 29060
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.271186440677966e-07,
+      "loss": 0.0,
+      "step": 29070
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.2697126013264554e-07,
+      "loss": 0.0003,
+      "step": 29080
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.2682387619749446e-07,
+      "loss": 0.0599,
+      "step": 29090
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.266764922623434e-07,
+      "loss": 0.2069,
+      "step": 29100
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.265291083271923e-07,
+      "loss": 0.0039,
+      "step": 29110
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.2638172439204126e-07,
+      "loss": 0.0001,
+      "step": 29120
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.262343404568902e-07,
+      "loss": 0.1201,
+      "step": 29130
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.260869565217391e-07,
+      "loss": 0.0,
+      "step": 29140
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.2593957258658806e-07,
+      "loss": 0.0001,
+      "step": 29150
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.2579218865143699e-07,
+      "loss": 0.0,
+      "step": 29160
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.256448047162859e-07,
+      "loss": 0.0001,
+      "step": 29170
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.2549742078113483e-07,
+      "loss": 0.0,
+      "step": 29180
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.2535003684598378e-07,
+      "loss": 0.0001,
+      "step": 29190
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.252026529108327e-07,
+      "loss": 0.0061,
+      "step": 29200
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.2505526897568163e-07,
+      "loss": 0.0001,
+      "step": 29210
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.2490788504053058e-07,
+      "loss": 0.1923,
+      "step": 29220
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.247605011053795e-07,
+      "loss": 0.0001,
+      "step": 29230
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.2461311717022846e-07,
+      "loss": 0.0001,
+      "step": 29240
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.2446573323507738e-07,
+      "loss": 0.2208,
+      "step": 29250
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.243183492999263e-07,
+      "loss": 0.4635,
+      "step": 29260
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.2417096536477523e-07,
+      "loss": 0.0001,
+      "step": 29270
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.2402358142962418e-07,
+      "loss": 0.0001,
+      "step": 29280
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.238761974944731e-07,
+      "loss": 0.0001,
+      "step": 29290
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.2372881355932203e-07,
+      "loss": 0.0001,
+      "step": 29300
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.2358142962417098e-07,
+      "loss": 0.1845,
+      "step": 29310
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.234340456890199e-07,
+      "loss": 0.0001,
+      "step": 29320
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.2328666175386882e-07,
+      "loss": 0.0001,
+      "step": 29330
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.2313927781871775e-07,
+      "loss": 0.0001,
+      "step": 29340
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.229918938835667e-07,
+      "loss": 0.0001,
+      "step": 29350
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.2284450994841562e-07,
+      "loss": 0.4026,
+      "step": 29360
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.2269712601326455e-07,
+      "loss": 0.0001,
+      "step": 29370
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.225497420781135e-07,
+      "loss": 0.0004,
+      "step": 29380
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.2240235814296242e-07,
+      "loss": 0.2122,
+      "step": 29390
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.2225497420781134e-07,
+      "loss": 0.0,
+      "step": 29400
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.2210759027266027e-07,
+      "loss": 0.1731,
+      "step": 29410
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.2196020633750922e-07,
+      "loss": 0.1478,
+      "step": 29420
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.2181282240235814e-07,
+      "loss": 0.0038,
+      "step": 29430
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.2166543846720707e-07,
+      "loss": 0.1561,
+      "step": 29440
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.2151805453205602e-07,
+      "loss": 0.1736,
+      "step": 29450
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.2137067059690494e-07,
+      "loss": 0.0009,
+      "step": 29460
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.2122328666175387e-07,
+      "loss": 0.0023,
+      "step": 29470
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.210759027266028e-07,
+      "loss": 0.0,
+      "step": 29480
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.2092851879145174e-07,
+      "loss": 0.0001,
+      "step": 29490
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.2078113485630066e-07,
+      "loss": 0.3281,
+      "step": 29500
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.206337509211496e-07,
+      "loss": 0.0001,
+      "step": 29510
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.2048636698599854e-07,
+      "loss": 0.2203,
+      "step": 29520
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.2033898305084746e-07,
+      "loss": 0.0001,
+      "step": 29530
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.2019159911569639e-07,
+      "loss": 0.0005,
+      "step": 29540
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.200442151805453e-07,
+      "loss": 0.0001,
+      "step": 29550
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.1989683124539426e-07,
+      "loss": 0.0,
+      "step": 29560
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.1974944731024318e-07,
+      "loss": 0.0001,
+      "step": 29570
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.196020633750921e-07,
+      "loss": 0.0002,
+      "step": 29580
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.1945467943994103e-07,
+      "loss": 0.0001,
+      "step": 29590
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.1930729550478998e-07,
+      "loss": 0.2513,
+      "step": 29600
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.191599115696389e-07,
+      "loss": 0.0001,
+      "step": 29610
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.1901252763448784e-07,
+      "loss": 0.0001,
+      "step": 29620
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.1886514369933677e-07,
+      "loss": 0.0,
+      "step": 29630
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.187177597641857e-07,
+      "loss": 0.0001,
+      "step": 29640
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.1857037582903463e-07,
+      "loss": 0.2804,
+      "step": 29650
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.1842299189388357e-07,
+      "loss": 0.0,
+      "step": 29660
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.1827560795873249e-07,
+      "loss": 0.0796,
+      "step": 29670
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.1812822402358143e-07,
+      "loss": 0.0001,
+      "step": 29680
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.1798084008843036e-07,
+      "loss": 0.1742,
+      "step": 29690
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.1783345615327929e-07,
+      "loss": 0.0001,
+      "step": 29700
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.1768607221812822e-07,
+      "loss": 0.5699,
+      "step": 29710
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.1753868828297715e-07,
+      "loss": 0.0001,
+      "step": 29720
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.1739130434782609e-07,
+      "loss": 0.0562,
+      "step": 29730
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.1724392041267501e-07,
+      "loss": 0.0001,
+      "step": 29740
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.1709653647752395e-07,
+      "loss": 0.1844,
+      "step": 29750
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.1694915254237288e-07,
+      "loss": 0.1391,
+      "step": 29760
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.1680176860722181e-07,
+      "loss": 0.0241,
+      "step": 29770
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.1665438467207074e-07,
+      "loss": 0.0,
+      "step": 29780
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.1650700073691967e-07,
+      "loss": 0.0065,
+      "step": 29790
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.163596168017686e-07,
+      "loss": 0.0004,
+      "step": 29800
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.1621223286661753e-07,
+      "loss": 0.0001,
+      "step": 29810
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.1606484893146647e-07,
+      "loss": 0.0001,
+      "step": 29820
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.159174649963154e-07,
+      "loss": 0.0001,
+      "step": 29830
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.1577008106116433e-07,
+      "loss": 0.0001,
+      "step": 29840
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.1562269712601327e-07,
+      "loss": 0.0001,
+      "step": 29850
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.1547531319086219e-07,
+      "loss": 0.0001,
+      "step": 29860
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.1532792925571113e-07,
+      "loss": 0.0022,
+      "step": 29870
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.1518054532056005e-07,
+      "loss": 0.0001,
+      "step": 29880
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.1503316138540899e-07,
+      "loss": 0.0284,
+      "step": 29890
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.1488577745025791e-07,
+      "loss": 0.0001,
+      "step": 29900
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.1473839351510685e-07,
+      "loss": 0.0001,
+      "step": 29910
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.1459100957995579e-07,
+      "loss": 0.0017,
+      "step": 29920
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.1444362564480471e-07,
+      "loss": 0.0001,
+      "step": 29930
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.1429624170965365e-07,
+      "loss": 0.0,
+      "step": 29940
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.1414885777450257e-07,
+      "loss": 0.4606,
+      "step": 29950
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.1400147383935151e-07,
+      "loss": 0.1123,
+      "step": 29960
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.1385408990420043e-07,
+      "loss": 0.0001,
+      "step": 29970
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.1370670596904937e-07,
+      "loss": 0.0001,
+      "step": 29980
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.135593220338983e-07,
+      "loss": 0.0001,
+      "step": 29990
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.1341193809874723e-07,
+      "loss": 0.196,
+      "step": 30000
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.1326455416359617e-07,
+      "loss": 0.0002,
+      "step": 30010
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.1311717022844509e-07,
+      "loss": 0.1215,
+      "step": 30020
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.1296978629329403e-07,
+      "loss": 0.0002,
+      "step": 30030
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.1282240235814295e-07,
+      "loss": 0.0828,
+      "step": 30040
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.1267501842299189e-07,
+      "loss": 0.0,
+      "step": 30050
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.1252763448784081e-07,
+      "loss": 0.1431,
+      "step": 30060
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.1238025055268975e-07,
+      "loss": 0.0,
+      "step": 30070
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.1223286661753869e-07,
+      "loss": 0.0001,
+      "step": 30080
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.1208548268238761e-07,
+      "loss": 0.0053,
+      "step": 30090
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.1193809874723655e-07,
+      "loss": 0.0001,
+      "step": 30100
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 1.1179071481208547e-07,
+      "loss": 0.0,
+      "step": 30110
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 1.1164333087693441e-07,
+      "loss": 0.0,
+      "step": 30120
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 1.1149594694178333e-07,
+      "loss": 0.0001,
+      "step": 30130
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 1.1134856300663227e-07,
+      "loss": 0.0218,
+      "step": 30140
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 1.1120117907148121e-07,
+      "loss": 0.0004,
+      "step": 30150
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 1.1105379513633013e-07,
+      "loss": 0.1943,
+      "step": 30160
+    },
+    {
+      "epoch": 11.07,
+      "eval_accuracy": 0.8988631909055272,
+      "eval_loss": 0.7849196195602417,
+      "eval_runtime": 617.7046,
+      "eval_samples_per_second": 4.13,
+      "eval_steps_per_second": 1.033,
+      "step": 30168
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.1090641120117907e-07,
+      "loss": 0.1778,
+      "step": 30170
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.1075902726602799e-07,
+      "loss": 0.0001,
+      "step": 30180
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.1061164333087693e-07,
+      "loss": 0.0001,
+      "step": 30190
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.1046425939572585e-07,
+      "loss": 0.0001,
+      "step": 30200
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.1031687546057479e-07,
+      "loss": 0.0001,
+      "step": 30210
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.1016949152542372e-07,
+      "loss": 0.0002,
+      "step": 30220
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.1002210759027265e-07,
+      "loss": 0.0001,
+      "step": 30230
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.0987472365512159e-07,
+      "loss": 0.0012,
+      "step": 30240
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.0972733971997051e-07,
+      "loss": 0.0002,
+      "step": 30250
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.0957995578481945e-07,
+      "loss": 0.0001,
+      "step": 30260
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.0943257184966837e-07,
+      "loss": 0.0001,
+      "step": 30270
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.0928518791451731e-07,
+      "loss": 0.0003,
+      "step": 30280
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.0913780397936624e-07,
+      "loss": 0.0001,
+      "step": 30290
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.0899042004421517e-07,
+      "loss": 0.0001,
+      "step": 30300
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.0884303610906411e-07,
+      "loss": 0.0001,
+      "step": 30310
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.0869565217391303e-07,
+      "loss": 0.0006,
+      "step": 30320
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.0854826823876197e-07,
+      "loss": 0.0001,
+      "step": 30330
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.084008843036109e-07,
+      "loss": 0.0001,
+      "step": 30340
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.0825350036845983e-07,
+      "loss": 0.0,
+      "step": 30350
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.0810611643330876e-07,
+      "loss": 0.0,
+      "step": 30360
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.0795873249815769e-07,
+      "loss": 0.0001,
+      "step": 30370
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.0781134856300663e-07,
+      "loss": 0.0952,
+      "step": 30380
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.0766396462785555e-07,
+      "loss": 0.0001,
+      "step": 30390
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.0751658069270449e-07,
+      "loss": 0.0002,
+      "step": 30400
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.0736919675755342e-07,
+      "loss": 0.0001,
+      "step": 30410
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.0722181282240235e-07,
+      "loss": 0.0001,
+      "step": 30420
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.0707442888725128e-07,
+      "loss": 0.0023,
+      "step": 30430
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.0692704495210021e-07,
+      "loss": 0.0002,
+      "step": 30440
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.0677966101694914e-07,
+      "loss": 0.0,
+      "step": 30450
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.0663227708179807e-07,
+      "loss": 0.0001,
+      "step": 30460
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.0648489314664701e-07,
+      "loss": 0.0,
+      "step": 30470
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.0633750921149594e-07,
+      "loss": 0.0134,
+      "step": 30480
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.0619012527634487e-07,
+      "loss": 0.2199,
+      "step": 30490
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.060427413411938e-07,
+      "loss": 0.0001,
+      "step": 30500
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.0589535740604273e-07,
+      "loss": 0.0001,
+      "step": 30510
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.0574797347089166e-07,
+      "loss": 0.0001,
+      "step": 30520
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.056005895357406e-07,
+      "loss": 0.0002,
+      "step": 30530
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.0545320560058953e-07,
+      "loss": 0.1469,
+      "step": 30540
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.0530582166543846e-07,
+      "loss": 0.0005,
+      "step": 30550
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.051584377302874e-07,
+      "loss": 0.0001,
+      "step": 30560
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.0501105379513632e-07,
+      "loss": 0.0001,
+      "step": 30570
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.0486366985998525e-07,
+      "loss": 0.0,
+      "step": 30580
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.0471628592483418e-07,
+      "loss": 0.0002,
+      "step": 30590
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.0456890198968312e-07,
+      "loss": 0.0001,
+      "step": 30600
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.0442151805453204e-07,
+      "loss": 0.0,
+      "step": 30610
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.0427413411938098e-07,
+      "loss": 0.139,
+      "step": 30620
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.0412675018422991e-07,
+      "loss": 0.0,
+      "step": 30630
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.0397936624907884e-07,
+      "loss": 0.0001,
+      "step": 30640
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.0383198231392777e-07,
+      "loss": 0.0001,
+      "step": 30650
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.036845983787767e-07,
+      "loss": 0.0142,
+      "step": 30660
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.0353721444362564e-07,
+      "loss": 0.0001,
+      "step": 30670
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.0338983050847456e-07,
+      "loss": 0.0001,
+      "step": 30680
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.032424465733235e-07,
+      "loss": 0.0,
+      "step": 30690
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.0309506263817243e-07,
+      "loss": 0.0563,
+      "step": 30700
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.0294767870302136e-07,
+      "loss": 0.0,
+      "step": 30710
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.028002947678703e-07,
+      "loss": 0.0001,
+      "step": 30720
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.0265291083271922e-07,
+      "loss": 0.0,
+      "step": 30730
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.0250552689756816e-07,
+      "loss": 0.0001,
+      "step": 30740
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.0235814296241708e-07,
+      "loss": 0.3791,
+      "step": 30750
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.0221075902726603e-07,
+      "loss": 0.0001,
+      "step": 30760
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.0206337509211497e-07,
+      "loss": 0.0,
+      "step": 30770
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.0191599115696389e-07,
+      "loss": 0.4283,
+      "step": 30780
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.0176860722181283e-07,
+      "loss": 0.0001,
+      "step": 30790
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.0162122328666175e-07,
+      "loss": 0.0001,
+      "step": 30800
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.0147383935151069e-07,
+      "loss": 0.0001,
+      "step": 30810
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.0132645541635961e-07,
+      "loss": 0.0,
+      "step": 30820
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.0117907148120855e-07,
+      "loss": 0.001,
+      "step": 30830
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.0103168754605748e-07,
+      "loss": 0.0001,
+      "step": 30840
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.0088430361090641e-07,
+      "loss": 0.0002,
+      "step": 30850
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.0073691967575535e-07,
+      "loss": 0.1845,
+      "step": 30860
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.0058953574060427e-07,
+      "loss": 0.0006,
+      "step": 30870
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.0044215180545321e-07,
+      "loss": 0.0,
+      "step": 30880
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.0029476787030213e-07,
+      "loss": 0.0006,
+      "step": 30890
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.0014738393515107e-07,
+      "loss": 0.0001,
+      "step": 30900
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1e-07,
+      "loss": 0.0001,
+      "step": 30910
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 9.985261606484893e-08,
+      "loss": 0.1238,
+      "step": 30920
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 9.970523212969787e-08,
+      "loss": 0.0001,
+      "step": 30930
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 9.95578481945468e-08,
+      "loss": 0.001,
+      "step": 30940
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 9.941046425939573e-08,
+      "loss": 0.0001,
+      "step": 30950
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 9.926308032424465e-08,
+      "loss": 0.0002,
+      "step": 30960
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 9.911569638909359e-08,
+      "loss": 0.0001,
+      "step": 30970
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 9.896831245394252e-08,
+      "loss": 0.0,
+      "step": 30980
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 9.882092851879145e-08,
+      "loss": 0.0001,
+      "step": 30990
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 9.867354458364038e-08,
+      "loss": 0.0005,
+      "step": 31000
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 9.852616064848931e-08,
+      "loss": 0.0,
+      "step": 31010
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 9.837877671333825e-08,
+      "loss": 0.0003,
+      "step": 31020
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 9.823139277818718e-08,
+      "loss": 0.0,
+      "step": 31030
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 9.808400884303611e-08,
+      "loss": 0.2119,
+      "step": 31040
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 9.793662490788504e-08,
+      "loss": 0.0001,
+      "step": 31050
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 9.778924097273397e-08,
+      "loss": 0.1604,
+      "step": 31060
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 9.76418570375829e-08,
+      "loss": 0.0,
+      "step": 31070
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 9.749447310243183e-08,
+      "loss": 0.0001,
+      "step": 31080
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 9.734708916728077e-08,
+      "loss": 0.0013,
+      "step": 31090
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 9.71997052321297e-08,
+      "loss": 0.0774,
+      "step": 31100
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 9.705232129697863e-08,
+      "loss": 0.0001,
+      "step": 31110
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.690493736182756e-08,
+      "loss": 0.0,
+      "step": 31120
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.67575534266765e-08,
+      "loss": 0.2274,
+      "step": 31130
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.661016949152542e-08,
+      "loss": 0.0031,
+      "step": 31140
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.646278555637435e-08,
+      "loss": 0.0338,
+      "step": 31150
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.631540162122329e-08,
+      "loss": 0.2746,
+      "step": 31160
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.616801768607222e-08,
+      "loss": 0.0001,
+      "step": 31170
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.602063375092115e-08,
+      "loss": 0.0,
+      "step": 31180
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.587324981577008e-08,
+      "loss": 0.0001,
+      "step": 31190
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.572586588061901e-08,
+      "loss": 0.0002,
+      "step": 31200
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.557848194546794e-08,
+      "loss": 0.0001,
+      "step": 31210
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.543109801031688e-08,
+      "loss": 0.0,
+      "step": 31220
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.52837140751658e-08,
+      "loss": 0.0,
+      "step": 31230
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.513633014001474e-08,
+      "loss": 0.0895,
+      "step": 31240
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.498894620486367e-08,
+      "loss": 0.0001,
+      "step": 31250
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.48415622697126e-08,
+      "loss": 0.0,
+      "step": 31260
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.469417833456153e-08,
+      "loss": 0.0006,
+      "step": 31270
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.454679439941046e-08,
+      "loss": 0.2261,
+      "step": 31280
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.43994104642594e-08,
+      "loss": 0.1593,
+      "step": 31290
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.425202652910832e-08,
+      "loss": 0.0001,
+      "step": 31300
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.410464259395726e-08,
+      "loss": 0.0,
+      "step": 31310
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.39572586588062e-08,
+      "loss": 0.0002,
+      "step": 31320
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.380987472365512e-08,
+      "loss": 0.0001,
+      "step": 31330
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.366249078850406e-08,
+      "loss": 0.3876,
+      "step": 31340
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.351510685335298e-08,
+      "loss": 0.0001,
+      "step": 31350
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.336772291820192e-08,
+      "loss": 0.0001,
+      "step": 31360
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.322033898305084e-08,
+      "loss": 0.0001,
+      "step": 31370
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.307295504789978e-08,
+      "loss": 0.0002,
+      "step": 31380
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.29255711127487e-08,
+      "loss": 0.1753,
+      "step": 31390
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.277818717759764e-08,
+      "loss": 0.0001,
+      "step": 31400
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.263080324244658e-08,
+      "loss": 0.0805,
+      "step": 31410
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.24834193072955e-08,
+      "loss": 0.0235,
+      "step": 31420
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.233603537214444e-08,
+      "loss": 0.066,
+      "step": 31430
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.218865143699336e-08,
+      "loss": 0.2133,
+      "step": 31440
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.20412675018423e-08,
+      "loss": 0.3465,
+      "step": 31450
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.189388356669122e-08,
+      "loss": 0.0051,
+      "step": 31460
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.174649963154016e-08,
+      "loss": 0.0002,
+      "step": 31470
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 9.15991156963891e-08,
+      "loss": 0.0001,
+      "step": 31480
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 9.145173176123802e-08,
+      "loss": 0.0547,
+      "step": 31490
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 9.130434782608696e-08,
+      "loss": 0.1754,
+      "step": 31500
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 9.115696389093588e-08,
+      "loss": 0.0,
+      "step": 31510
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 9.100957995578482e-08,
+      "loss": 0.0,
+      "step": 31520
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 9.086219602063374e-08,
+      "loss": 0.0,
+      "step": 31530
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 9.071481208548268e-08,
+      "loss": 0.0001,
+      "step": 31540
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 9.05674281503316e-08,
+      "loss": 0.0572,
+      "step": 31550
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 9.042004421518054e-08,
+      "loss": 0.0012,
+      "step": 31560
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 9.027266028002948e-08,
+      "loss": 0.0,
+      "step": 31570
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 9.01252763448784e-08,
+      "loss": 0.0,
+      "step": 31580
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.997789240972734e-08,
+      "loss": 0.0,
+      "step": 31590
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.983050847457626e-08,
+      "loss": 0.0,
+      "step": 31600
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.96831245394252e-08,
+      "loss": 0.0,
+      "step": 31610
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.953574060427412e-08,
+      "loss": 0.0001,
+      "step": 31620
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.938835666912306e-08,
+      "loss": 0.1094,
+      "step": 31630
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.9240972733972e-08,
+      "loss": 0.0001,
+      "step": 31640
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.909358879882092e-08,
+      "loss": 0.0,
+      "step": 31650
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.894620486366986e-08,
+      "loss": 0.0001,
+      "step": 31660
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.879882092851878e-08,
+      "loss": 0.0001,
+      "step": 31670
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.865143699336772e-08,
+      "loss": 0.0,
+      "step": 31680
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.850405305821664e-08,
+      "loss": 0.0001,
+      "step": 31690
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.835666912306558e-08,
+      "loss": 0.0001,
+      "step": 31700
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.820928518791452e-08,
+      "loss": 0.0,
+      "step": 31710
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.806190125276344e-08,
+      "loss": 0.0001,
+      "step": 31720
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.791451731761238e-08,
+      "loss": 0.0,
+      "step": 31730
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.77671333824613e-08,
+      "loss": 0.0003,
+      "step": 31740
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.761974944731024e-08,
+      "loss": 0.0001,
+      "step": 31750
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.747236551215916e-08,
+      "loss": 0.0,
+      "step": 31760
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.73249815770081e-08,
+      "loss": 0.1841,
+      "step": 31770
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.717759764185703e-08,
+      "loss": 0.1978,
+      "step": 31780
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.703021370670596e-08,
+      "loss": 0.1441,
+      "step": 31790
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.68828297715549e-08,
+      "loss": 0.0098,
+      "step": 31800
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.673544583640382e-08,
+      "loss": 0.2049,
+      "step": 31810
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.658806190125276e-08,
+      "loss": 0.0019,
+      "step": 31820
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.644067796610168e-08,
+      "loss": 0.0001,
+      "step": 31830
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.629329403095062e-08,
+      "loss": 0.0,
+      "step": 31840
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.614591009579955e-08,
+      "loss": 0.0001,
+      "step": 31850
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.599852616064848e-08,
+      "loss": 0.0001,
+      "step": 31860
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.585114222549742e-08,
+      "loss": 0.0001,
+      "step": 31870
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.570375829034634e-08,
+      "loss": 0.0001,
+      "step": 31880
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.555637435519528e-08,
+      "loss": 0.0001,
+      "step": 31890
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.54089904200442e-08,
+      "loss": 0.0001,
+      "step": 31900
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.526160648489314e-08,
+      "loss": 0.0001,
+      "step": 31910
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.511422254974207e-08,
+      "loss": 0.3625,
+      "step": 31920
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.4966838614591e-08,
+      "loss": 0.5146,
+      "step": 31930
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.481945467943993e-08,
+      "loss": 0.0001,
+      "step": 31940
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.467207074428886e-08,
+      "loss": 0.0,
+      "step": 31950
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.45246868091378e-08,
+      "loss": 0.1594,
+      "step": 31960
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.437730287398673e-08,
+      "loss": 0.0001,
+      "step": 31970
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.422991893883566e-08,
+      "loss": 0.0001,
+      "step": 31980
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.408253500368459e-08,
+      "loss": 0.0002,
+      "step": 31990
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.393515106853352e-08,
+      "loss": 0.0001,
+      "step": 32000
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.378776713338245e-08,
+      "loss": 0.0,
+      "step": 32010
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.364038319823138e-08,
+      "loss": 0.0001,
+      "step": 32020
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.349299926308032e-08,
+      "loss": 0.0,
+      "step": 32030
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.334561532792925e-08,
+      "loss": 0.0001,
+      "step": 32040
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.319823139277818e-08,
+      "loss": 0.027,
+      "step": 32050
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.305084745762711e-08,
+      "loss": 0.0001,
+      "step": 32060
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.290346352247604e-08,
+      "loss": 0.0,
+      "step": 32070
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.275607958732497e-08,
+      "loss": 0.0,
+      "step": 32080
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.26086956521739e-08,
+      "loss": 0.0,
+      "step": 32090
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.246131171702283e-08,
+      "loss": 0.0,
+      "step": 32100
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.231392778187177e-08,
+      "loss": 0.0001,
+      "step": 32110
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.21665438467207e-08,
+      "loss": 0.0,
+      "step": 32120
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.201915991156963e-08,
+      "loss": 0.0001,
+      "step": 32130
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.187177597641856e-08,
+      "loss": 0.0001,
+      "step": 32140
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.172439204126749e-08,
+      "loss": 0.0,
+      "step": 32150
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.157700810611643e-08,
+      "loss": 0.0001,
+      "step": 32160
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.142962417096535e-08,
+      "loss": 0.16,
+      "step": 32170
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.128224023581429e-08,
+      "loss": 0.0002,
+      "step": 32180
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.113485630066322e-08,
+      "loss": 0.2197,
+      "step": 32190
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.098747236551215e-08,
+      "loss": 0.1691,
+      "step": 32200
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.084008843036109e-08,
+      "loss": 0.184,
+      "step": 32210
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.069270449521001e-08,
+      "loss": 0.0,
+      "step": 32220
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.054532056005895e-08,
+      "loss": 0.0011,
+      "step": 32230
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.039793662490787e-08,
+      "loss": 0.0002,
+      "step": 32240
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 8.025055268975681e-08,
+      "loss": 0.0001,
+      "step": 32250
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 8.010316875460574e-08,
+      "loss": 0.1826,
+      "step": 32260
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.995578481945467e-08,
+      "loss": 0.0806,
+      "step": 32270
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.98084008843036e-08,
+      "loss": 0.1593,
+      "step": 32280
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.966101694915253e-08,
+      "loss": 0.0004,
+      "step": 32290
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.951363301400148e-08,
+      "loss": 0.0001,
+      "step": 32300
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.93662490788504e-08,
+      "loss": 0.0001,
+      "step": 32310
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.921886514369934e-08,
+      "loss": 0.0001,
+      "step": 32320
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.907148120854826e-08,
+      "loss": 0.0001,
+      "step": 32330
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.89240972733972e-08,
+      "loss": 0.0001,
+      "step": 32340
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.877671333824614e-08,
+      "loss": 0.0001,
+      "step": 32350
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.862932940309506e-08,
+      "loss": 0.0001,
+      "step": 32360
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.8481945467944e-08,
+      "loss": 0.0,
+      "step": 32370
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.833456153279292e-08,
+      "loss": 0.0,
+      "step": 32380
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.818717759764186e-08,
+      "loss": 0.0001,
+      "step": 32390
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.803979366249079e-08,
+      "loss": 0.0001,
+      "step": 32400
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.789240972733972e-08,
+      "loss": 0.0,
+      "step": 32410
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.774502579218866e-08,
+      "loss": 0.0008,
+      "step": 32420
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.759764185703758e-08,
+      "loss": 0.1429,
+      "step": 32430
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.745025792188652e-08,
+      "loss": 0.0001,
+      "step": 32440
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.730287398673544e-08,
+      "loss": 0.0002,
+      "step": 32450
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.715549005158438e-08,
+      "loss": 0.0,
+      "step": 32460
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.70081061164333e-08,
+      "loss": 0.0026,
+      "step": 32470
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.686072218128224e-08,
+      "loss": 0.0002,
+      "step": 32480
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.671333824613118e-08,
+      "loss": 0.0001,
+      "step": 32490
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.65659543109801e-08,
+      "loss": 0.0001,
+      "step": 32500
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.641857037582904e-08,
+      "loss": 0.0001,
+      "step": 32510
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.627118644067796e-08,
+      "loss": 0.076,
+      "step": 32520
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.61238025055269e-08,
+      "loss": 0.0001,
+      "step": 32530
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.597641857037583e-08,
+      "loss": 0.0001,
+      "step": 32540
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.582903463522476e-08,
+      "loss": 0.214,
+      "step": 32550
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.568165070007369e-08,
+      "loss": 0.0001,
+      "step": 32560
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.553426676492262e-08,
+      "loss": 0.0,
+      "step": 32570
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.538688282977156e-08,
+      "loss": 0.0001,
+      "step": 32580
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.523949889462049e-08,
+      "loss": 0.0001,
+      "step": 32590
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.509211495946942e-08,
+      "loss": 0.0001,
+      "step": 32600
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 7.494473102431835e-08,
+      "loss": 0.0001,
+      "step": 32610
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 7.479734708916728e-08,
+      "loss": 0.2756,
+      "step": 32620
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 7.464996315401621e-08,
+      "loss": 0.0,
+      "step": 32630
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 7.450257921886514e-08,
+      "loss": 0.0003,
+      "step": 32640
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 7.435519528371408e-08,
+      "loss": 0.0001,
+      "step": 32650
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 7.4207811348563e-08,
+      "loss": 0.2704,
+      "step": 32660
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 7.406042741341194e-08,
+      "loss": 0.1019,
+      "step": 32670
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 7.391304347826087e-08,
+      "loss": 0.0001,
+      "step": 32680
+    },
+    {
+      "epoch": 12.07,
+      "eval_accuracy": 0.898079184633477,
+      "eval_loss": 0.8075343370437622,
+      "eval_runtime": 596.6043,
+      "eval_samples_per_second": 4.276,
+      "eval_steps_per_second": 1.069,
+      "step": 32682
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 7.37656595431098e-08,
+      "loss": 0.0001,
+      "step": 32690
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 7.361827560795873e-08,
+      "loss": 0.0,
+      "step": 32700
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 7.347089167280767e-08,
+      "loss": 0.0001,
+      "step": 32710
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 7.332350773765659e-08,
+      "loss": 0.0001,
+      "step": 32720
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 7.317612380250553e-08,
+      "loss": 0.0001,
+      "step": 32730
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 7.302873986735446e-08,
+      "loss": 0.0008,
+      "step": 32740
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 7.288135593220339e-08,
+      "loss": 0.154,
+      "step": 32750
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 7.273397199705232e-08,
+      "loss": 0.0,
+      "step": 32760
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 7.258658806190125e-08,
+      "loss": 0.0,
+      "step": 32770
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 7.243920412675019e-08,
+      "loss": 0.0001,
+      "step": 32780
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 7.229182019159911e-08,
+      "loss": 0.1624,
+      "step": 32790
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 7.214443625644805e-08,
+      "loss": 0.2029,
+      "step": 32800
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 7.199705232129698e-08,
+      "loss": 0.0001,
+      "step": 32810
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 7.184966838614591e-08,
+      "loss": 0.0001,
+      "step": 32820
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 7.170228445099484e-08,
+      "loss": 0.1123,
+      "step": 32830
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 7.155490051584377e-08,
+      "loss": 0.0001,
+      "step": 32840
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 7.14075165806927e-08,
+      "loss": 0.0001,
+      "step": 32850
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 7.126013264554163e-08,
+      "loss": 0.0001,
+      "step": 32860
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 7.111274871039057e-08,
+      "loss": 0.0,
+      "step": 32870
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 7.096536477523949e-08,
+      "loss": 0.0001,
+      "step": 32880
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 7.081798084008843e-08,
+      "loss": 0.0001,
+      "step": 32890
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 7.067059690493737e-08,
+      "loss": 0.0003,
+      "step": 32900
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 7.052321296978629e-08,
+      "loss": 0.0,
+      "step": 32910
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 7.037582903463523e-08,
+      "loss": 0.0006,
+      "step": 32920
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 7.022844509948415e-08,
+      "loss": 0.0081,
+      "step": 32930
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 7.008106116433309e-08,
+      "loss": 0.0001,
+      "step": 32940
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 6.993367722918201e-08,
+      "loss": 0.0,
+      "step": 32950
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 6.978629329403095e-08,
+      "loss": 0.1638,
+      "step": 32960
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 6.963890935887989e-08,
+      "loss": 0.0001,
+      "step": 32970
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 6.949152542372881e-08,
+      "loss": 0.1719,
+      "step": 32980
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 6.934414148857775e-08,
+      "loss": 0.0001,
+      "step": 32990
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 6.919675755342667e-08,
+      "loss": 0.0001,
+      "step": 33000
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 6.904937361827561e-08,
+      "loss": 0.0002,
+      "step": 33010
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 6.890198968312453e-08,
+      "loss": 0.1668,
+      "step": 33020
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 6.875460574797347e-08,
+      "loss": 0.0,
+      "step": 33030
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 6.86072218128224e-08,
+      "loss": 0.0,
+      "step": 33040
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 6.845983787767133e-08,
+      "loss": 0.0003,
+      "step": 33050
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 6.831245394252027e-08,
+      "loss": 0.0003,
+      "step": 33060
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 6.816507000736919e-08,
+      "loss": 0.0,
+      "step": 33070
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 6.801768607221813e-08,
+      "loss": 0.0001,
+      "step": 33080
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 6.787030213706705e-08,
+      "loss": 0.0001,
+      "step": 33090
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 6.772291820191599e-08,
+      "loss": 0.0003,
+      "step": 33100
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 6.757553426676491e-08,
+      "loss": 0.0,
+      "step": 33110
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 6.742815033161385e-08,
+      "loss": 0.0,
+      "step": 33120
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 6.728076639646279e-08,
+      "loss": 0.0,
+      "step": 33130
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 6.713338246131171e-08,
+      "loss": 0.0,
+      "step": 33140
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 6.698599852616065e-08,
+      "loss": 0.1417,
+      "step": 33150
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 6.683861459100957e-08,
+      "loss": 0.0072,
+      "step": 33160
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 6.669123065585851e-08,
+      "loss": 0.0006,
+      "step": 33170
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 6.654384672070743e-08,
+      "loss": 0.0,
+      "step": 33180
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 6.639646278555637e-08,
+      "loss": 0.0001,
+      "step": 33190
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 6.624907885040531e-08,
+      "loss": 0.0,
+      "step": 33200
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 6.610169491525423e-08,
+      "loss": 0.0001,
+      "step": 33210
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 6.595431098010317e-08,
+      "loss": 0.0,
+      "step": 33220
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 6.580692704495209e-08,
+      "loss": 0.1778,
+      "step": 33230
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 6.565954310980103e-08,
+      "loss": 0.1568,
+      "step": 33240
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.551215917464995e-08,
+      "loss": 0.0,
+      "step": 33250
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.536477523949889e-08,
+      "loss": 0.0001,
+      "step": 33260
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.521739130434782e-08,
+      "loss": 0.0001,
+      "step": 33270
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.507000736919675e-08,
+      "loss": 0.1585,
+      "step": 33280
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.492262343404569e-08,
+      "loss": 0.0443,
+      "step": 33290
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.477523949889461e-08,
+      "loss": 0.0001,
+      "step": 33300
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.462785556374355e-08,
+      "loss": 0.0,
+      "step": 33310
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.448047162859247e-08,
+      "loss": 0.1512,
+      "step": 33320
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.433308769344141e-08,
+      "loss": 0.0001,
+      "step": 33330
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.418570375829034e-08,
+      "loss": 0.0002,
+      "step": 33340
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.403831982313927e-08,
+      "loss": 0.1689,
+      "step": 33350
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.389093588798821e-08,
+      "loss": 0.0001,
+      "step": 33360
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.374355195283713e-08,
+      "loss": 0.0,
+      "step": 33370
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.359616801768607e-08,
+      "loss": 0.0,
+      "step": 33380
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.3448784082535e-08,
+      "loss": 0.0,
+      "step": 33390
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.330140014738393e-08,
+      "loss": 0.0001,
+      "step": 33400
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.315401621223286e-08,
+      "loss": 0.0,
+      "step": 33410
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.300663227708179e-08,
+      "loss": 0.0001,
+      "step": 33420
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.285924834193072e-08,
+      "loss": 0.0,
+      "step": 33430
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.271186440677965e-08,
+      "loss": 0.2241,
+      "step": 33440
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.256448047162859e-08,
+      "loss": 0.0035,
+      "step": 33450
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.241709653647753e-08,
+      "loss": 0.0001,
+      "step": 33460
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.226971260132645e-08,
+      "loss": 0.0001,
+      "step": 33470
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.212232866617539e-08,
+      "loss": 0.0905,
+      "step": 33480
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.197494473102431e-08,
+      "loss": 0.0001,
+      "step": 33490
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.182756079587325e-08,
+      "loss": 0.0001,
+      "step": 33500
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.168017686072219e-08,
+      "loss": 0.0002,
+      "step": 33510
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.153279292557111e-08,
+      "loss": 0.046,
+      "step": 33520
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.138540899042005e-08,
+      "loss": 0.0001,
+      "step": 33530
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.123802505526897e-08,
+      "loss": 0.0001,
+      "step": 33540
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.109064112011791e-08,
+      "loss": 0.0001,
+      "step": 33550
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.094325718496683e-08,
+      "loss": 0.0001,
+      "step": 33560
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.079587324981577e-08,
+      "loss": 0.0,
+      "step": 33570
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.06484893146647e-08,
+      "loss": 0.0001,
+      "step": 33580
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.050110537951363e-08,
+      "loss": 0.0,
+      "step": 33590
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.035372144436257e-08,
+      "loss": 0.0001,
+      "step": 33600
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.020633750921149e-08,
+      "loss": 0.0008,
+      "step": 33610
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 6.005895357406043e-08,
+      "loss": 0.0001,
+      "step": 33620
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.991156963890935e-08,
+      "loss": 0.0006,
+      "step": 33630
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.976418570375829e-08,
+      "loss": 0.0,
+      "step": 33640
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.961680176860722e-08,
+      "loss": 0.0,
+      "step": 33650
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.946941783345615e-08,
+      "loss": 0.066,
+      "step": 33660
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.932203389830508e-08,
+      "loss": 0.0,
+      "step": 33670
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.9174649963154014e-08,
+      "loss": 0.0001,
+      "step": 33680
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.9027266028002944e-08,
+      "loss": 0.0001,
+      "step": 33690
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.8879882092851875e-08,
+      "loss": 0.2026,
+      "step": 33700
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.873249815770081e-08,
+      "loss": 0.1869,
+      "step": 33710
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.858511422254974e-08,
+      "loss": 0.1555,
+      "step": 33720
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.843773028739867e-08,
+      "loss": 0.0577,
+      "step": 33730
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.8290346352247603e-08,
+      "loss": 0.0,
+      "step": 33740
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.8142962417096534e-08,
+      "loss": 0.0001,
+      "step": 33750
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.7995578481945465e-08,
+      "loss": 0.0001,
+      "step": 33760
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.7848194546794395e-08,
+      "loss": 0.0003,
+      "step": 33770
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.7700810611643326e-08,
+      "loss": 0.0001,
+      "step": 33780
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.755342667649226e-08,
+      "loss": 0.0001,
+      "step": 33790
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.7406042741341193e-08,
+      "loss": 0.0,
+      "step": 33800
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.7258658806190124e-08,
+      "loss": 0.0001,
+      "step": 33810
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.7111274871039054e-08,
+      "loss": 0.1657,
+      "step": 33820
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.6963890935887985e-08,
+      "loss": 0.0001,
+      "step": 33830
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.6816507000736916e-08,
+      "loss": 0.0,
+      "step": 33840
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.6669123065585846e-08,
+      "loss": 0.0004,
+      "step": 33850
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.6521739130434777e-08,
+      "loss": 0.0,
+      "step": 33860
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.6374355195283714e-08,
+      "loss": 0.0001,
+      "step": 33870
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.6226971260132644e-08,
+      "loss": 0.0001,
+      "step": 33880
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.6079587324981575e-08,
+      "loss": 0.1357,
+      "step": 33890
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.5932203389830505e-08,
+      "loss": 0.0441,
+      "step": 33900
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.5784819454679436e-08,
+      "loss": 0.0001,
+      "step": 33910
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.5637435519528366e-08,
+      "loss": 0.0001,
+      "step": 33920
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.54900515843773e-08,
+      "loss": 0.0806,
+      "step": 33930
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.5342667649226234e-08,
+      "loss": 0.0,
+      "step": 33940
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.5195283714075165e-08,
+      "loss": 0.0001,
+      "step": 33950
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.5047899778924095e-08,
+      "loss": 0.0047,
+      "step": 33960
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.4900515843773026e-08,
+      "loss": 0.0001,
+      "step": 33970
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.4753131908621956e-08,
+      "loss": 0.0079,
+      "step": 33980
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.460574797347089e-08,
+      "loss": 0.0,
+      "step": 33990
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.445836403831982e-08,
+      "loss": 0.0,
+      "step": 34000
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 5.431098010316875e-08,
+      "loss": 0.0,
+      "step": 34010
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 5.4163596168017685e-08,
+      "loss": 0.0,
+      "step": 34020
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 5.4016212232866616e-08,
+      "loss": 0.0001,
+      "step": 34030
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 5.3868828297715546e-08,
+      "loss": 0.0001,
+      "step": 34040
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 5.372144436256448e-08,
+      "loss": 0.0,
+      "step": 34050
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 5.357406042741341e-08,
+      "loss": 0.0003,
+      "step": 34060
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 5.342667649226234e-08,
+      "loss": 0.0002,
+      "step": 34070
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 5.327929255711127e-08,
+      "loss": 0.1107,
+      "step": 34080
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 5.31319086219602e-08,
+      "loss": 0.0,
+      "step": 34090
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 5.2984524686809136e-08,
+      "loss": 0.1,
+      "step": 34100
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 5.2837140751658067e-08,
+      "loss": 0.0,
+      "step": 34110
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 5.2689756816507e-08,
+      "loss": 0.0,
+      "step": 34120
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 5.254237288135593e-08,
+      "loss": 0.0,
+      "step": 34130
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 5.239498894620486e-08,
+      "loss": 0.0001,
+      "step": 34140
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 5.224760501105379e-08,
+      "loss": 0.0001,
+      "step": 34150
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 5.210022107590272e-08,
+      "loss": 0.0001,
+      "step": 34160
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 5.195283714075165e-08,
+      "loss": 0.0,
+      "step": 34170
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 5.180545320560059e-08,
+      "loss": 0.0,
+      "step": 34180
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 5.165806927044952e-08,
+      "loss": 0.1488,
+      "step": 34190
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 5.151068533529845e-08,
+      "loss": 0.0001,
+      "step": 34200
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 5.136330140014738e-08,
+      "loss": 0.0001,
+      "step": 34210
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 5.121591746499631e-08,
+      "loss": 0.0003,
+      "step": 34220
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 5.1068533529845246e-08,
+      "loss": 0.0001,
+      "step": 34230
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 5.092114959469418e-08,
+      "loss": 0.0003,
+      "step": 34240
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 5.0773765659543114e-08,
+      "loss": 0.0,
+      "step": 34250
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 5.0626381724392045e-08,
+      "loss": 0.0001,
+      "step": 34260
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 5.0478997789240975e-08,
+      "loss": 0.0001,
+      "step": 34270
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 5.0331613854089906e-08,
+      "loss": 0.0,
+      "step": 34280
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 5.0184229918938836e-08,
+      "loss": 0.2442,
+      "step": 34290
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 5.003684598378777e-08,
+      "loss": 0.0,
+      "step": 34300
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.98894620486367e-08,
+      "loss": 0.0001,
+      "step": 34310
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.974207811348563e-08,
+      "loss": 0.0,
+      "step": 34320
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.9594694178334565e-08,
+      "loss": 0.0,
+      "step": 34330
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.9447310243183496e-08,
+      "loss": 0.0527,
+      "step": 34340
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.9299926308032426e-08,
+      "loss": 0.0001,
+      "step": 34350
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.9152542372881357e-08,
+      "loss": 0.0003,
+      "step": 34360
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.900515843773029e-08,
+      "loss": 0.0001,
+      "step": 34370
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.885777450257922e-08,
+      "loss": 0.0001,
+      "step": 34380
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.871039056742815e-08,
+      "loss": 0.0001,
+      "step": 34390
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.856300663227708e-08,
+      "loss": 0.0001,
+      "step": 34400
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.8415622697126016e-08,
+      "loss": 0.0001,
+      "step": 34410
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.8268238761974946e-08,
+      "loss": 0.0001,
+      "step": 34420
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.812085482682388e-08,
+      "loss": 0.0001,
+      "step": 34430
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.797347089167281e-08,
+      "loss": 0.0001,
+      "step": 34440
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.782608695652174e-08,
+      "loss": 0.0,
+      "step": 34450
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.767870302137067e-08,
+      "loss": 0.0,
+      "step": 34460
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.75313190862196e-08,
+      "loss": 0.1938,
+      "step": 34470
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.738393515106853e-08,
+      "loss": 0.0,
+      "step": 34480
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.723655121591747e-08,
+      "loss": 0.0001,
+      "step": 34490
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.70891672807664e-08,
+      "loss": 0.0144,
+      "step": 34500
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.694178334561533e-08,
+      "loss": 0.0,
+      "step": 34510
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.679439941046426e-08,
+      "loss": 0.0001,
+      "step": 34520
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.664701547531319e-08,
+      "loss": 0.0001,
+      "step": 34530
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.649963154016212e-08,
+      "loss": 0.1187,
+      "step": 34540
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.635224760501105e-08,
+      "loss": 0.0,
+      "step": 34550
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.620486366985998e-08,
+      "loss": 0.0017,
+      "step": 34560
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.605747973470892e-08,
+      "loss": 0.1854,
+      "step": 34570
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.591009579955785e-08,
+      "loss": 0.0001,
+      "step": 34580
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.576271186440678e-08,
+      "loss": 0.022,
+      "step": 34590
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.561532792925571e-08,
+      "loss": 0.0,
+      "step": 34600
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.546794399410464e-08,
+      "loss": 0.0001,
+      "step": 34610
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.532056005895357e-08,
+      "loss": 0.0,
+      "step": 34620
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.51731761238025e-08,
+      "loss": 0.0001,
+      "step": 34630
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.502579218865143e-08,
+      "loss": 0.0058,
+      "step": 34640
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.487840825350037e-08,
+      "loss": 0.0,
+      "step": 34650
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.47310243183493e-08,
+      "loss": 0.206,
+      "step": 34660
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.458364038319823e-08,
+      "loss": 0.0001,
+      "step": 34670
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.443625644804716e-08,
+      "loss": 0.1332,
+      "step": 34680
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.428887251289609e-08,
+      "loss": 0.1996,
+      "step": 34690
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.414148857774502e-08,
+      "loss": 0.0001,
+      "step": 34700
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.399410464259395e-08,
+      "loss": 0.0072,
+      "step": 34710
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.384672070744288e-08,
+      "loss": 0.0001,
+      "step": 34720
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.369933677229182e-08,
+      "loss": 0.0,
+      "step": 34730
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.355195283714075e-08,
+      "loss": 0.0,
+      "step": 34740
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.340456890198968e-08,
+      "loss": 0.1203,
+      "step": 34750
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.325718496683861e-08,
+      "loss": 0.0,
+      "step": 34760
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.310980103168754e-08,
+      "loss": 0.0166,
+      "step": 34770
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.296241709653647e-08,
+      "loss": 0.0001,
+      "step": 34780
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.28150331613854e-08,
+      "loss": 0.0,
+      "step": 34790
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.2667649226234334e-08,
+      "loss": 0.0834,
+      "step": 34800
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.252026529108327e-08,
+      "loss": 0.154,
+      "step": 34810
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.23728813559322e-08,
+      "loss": 0.1701,
+      "step": 34820
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.222549742078113e-08,
+      "loss": 0.0,
+      "step": 34830
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.207811348563006e-08,
+      "loss": 0.0,
+      "step": 34840
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.193072955047899e-08,
+      "loss": 0.0002,
+      "step": 34850
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.1783345615327923e-08,
+      "loss": 0.0,
+      "step": 34860
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.1635961680176854e-08,
+      "loss": 0.1865,
+      "step": 34870
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.148857774502579e-08,
+      "loss": 0.0,
+      "step": 34880
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.134119380987472e-08,
+      "loss": 0.0,
+      "step": 34890
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.119380987472365e-08,
+      "loss": 0.0001,
+      "step": 34900
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.104642593957258e-08,
+      "loss": 0.0,
+      "step": 34910
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.089904200442151e-08,
+      "loss": 0.0455,
+      "step": 34920
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.0751658069270444e-08,
+      "loss": 0.0,
+      "step": 34930
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.0604274134119374e-08,
+      "loss": 0.0,
+      "step": 34940
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.0456890198968305e-08,
+      "loss": 0.0,
+      "step": 34950
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.030950626381724e-08,
+      "loss": 0.0,
+      "step": 34960
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.016212232866617e-08,
+      "loss": 0.0095,
+      "step": 34970
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.00147383935151e-08,
+      "loss": 0.0601,
+      "step": 34980
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 3.9867354458364034e-08,
+      "loss": 0.1628,
+      "step": 34990
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 3.971997052321297e-08,
+      "loss": 0.0001,
+      "step": 35000
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 3.95725865880619e-08,
+      "loss": 0.004,
+      "step": 35010
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 3.942520265291083e-08,
+      "loss": 0.0,
+      "step": 35020
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 3.927781871775976e-08,
+      "loss": 0.0,
+      "step": 35030
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 3.91304347826087e-08,
+      "loss": 0.0001,
+      "step": 35040
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 3.898305084745763e-08,
+      "loss": 0.0,
+      "step": 35050
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 3.883566691230656e-08,
+      "loss": 0.0001,
+      "step": 35060
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 3.868828297715549e-08,
+      "loss": 0.0001,
+      "step": 35070
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 3.854089904200442e-08,
+      "loss": 0.1903,
+      "step": 35080
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 3.839351510685335e-08,
+      "loss": 0.0,
+      "step": 35090
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 3.824613117170228e-08,
+      "loss": 0.0001,
+      "step": 35100
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 3.8098747236551213e-08,
+      "loss": 0.0,
+      "step": 35110
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 3.795136330140015e-08,
+      "loss": 0.0,
+      "step": 35120
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 3.780397936624908e-08,
+      "loss": 0.0,
+      "step": 35130
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 3.765659543109801e-08,
+      "loss": 0.1803,
+      "step": 35140
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 3.750921149594694e-08,
+      "loss": 0.2325,
+      "step": 35150
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 3.736182756079587e-08,
+      "loss": 0.0024,
+      "step": 35160
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 3.72144436256448e-08,
+      "loss": 0.0001,
+      "step": 35170
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 3.7067059690493734e-08,
+      "loss": 0.0,
+      "step": 35180
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 3.691967575534267e-08,
+      "loss": 0.0,
+      "step": 35190
+    },
+    {
+      "epoch": 13.07,
+      "eval_accuracy": 0.8961191689533516,
+      "eval_loss": 0.8346682786941528,
+      "eval_runtime": 561.4225,
+      "eval_samples_per_second": 4.544,
+      "eval_steps_per_second": 1.136,
+      "step": 35196
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.67722918201916e-08,
+      "loss": 0.1796,
+      "step": 35200
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.662490788504053e-08,
+      "loss": 0.0001,
+      "step": 35210
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.647752394988946e-08,
+      "loss": 0.0,
+      "step": 35220
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.633014001473839e-08,
+      "loss": 0.0001,
+      "step": 35230
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.6182756079587324e-08,
+      "loss": 0.0003,
+      "step": 35240
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.6035372144436254e-08,
+      "loss": 0.0,
+      "step": 35250
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.5887988209285185e-08,
+      "loss": 0.0022,
+      "step": 35260
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.574060427413412e-08,
+      "loss": 0.0001,
+      "step": 35270
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.559322033898305e-08,
+      "loss": 0.0,
+      "step": 35280
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.544583640383198e-08,
+      "loss": 0.0,
+      "step": 35290
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.5298452468680914e-08,
+      "loss": 0.0,
+      "step": 35300
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.5151068533529844e-08,
+      "loss": 0.0,
+      "step": 35310
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.5003684598378775e-08,
+      "loss": 0.0,
+      "step": 35320
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.4856300663227705e-08,
+      "loss": 0.0001,
+      "step": 35330
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.4708916728076636e-08,
+      "loss": 0.0,
+      "step": 35340
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.456153279292557e-08,
+      "loss": 0.0001,
+      "step": 35350
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.4414148857774503e-08,
+      "loss": 0.0,
+      "step": 35360
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.4266764922623434e-08,
+      "loss": 0.0,
+      "step": 35370
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.4119380987472365e-08,
+      "loss": 0.0002,
+      "step": 35380
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.3971997052321295e-08,
+      "loss": 0.0615,
+      "step": 35390
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.3824613117170226e-08,
+      "loss": 0.0007,
+      "step": 35400
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.3677229182019156e-08,
+      "loss": 0.0001,
+      "step": 35410
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.3529845246868087e-08,
+      "loss": 0.0001,
+      "step": 35420
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.3382461311717024e-08,
+      "loss": 0.0001,
+      "step": 35430
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.3235077376565954e-08,
+      "loss": 0.0,
+      "step": 35440
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.3087693441414885e-08,
+      "loss": 0.0004,
+      "step": 35450
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.2940309506263815e-08,
+      "loss": 0.0026,
+      "step": 35460
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.2792925571112746e-08,
+      "loss": 0.0151,
+      "step": 35470
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.2645541635961677e-08,
+      "loss": 0.0001,
+      "step": 35480
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.249815770081061e-08,
+      "loss": 0.0,
+      "step": 35490
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.235077376565954e-08,
+      "loss": 0.0001,
+      "step": 35500
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.2203389830508475e-08,
+      "loss": 0.103,
+      "step": 35510
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.2056005895357405e-08,
+      "loss": 0.0,
+      "step": 35520
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.1908621960206336e-08,
+      "loss": 0.1594,
+      "step": 35530
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.1761238025055266e-08,
+      "loss": 0.0,
+      "step": 35540
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.16138540899042e-08,
+      "loss": 0.0,
+      "step": 35550
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.146647015475313e-08,
+      "loss": 0.0,
+      "step": 35560
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.131908621960206e-08,
+      "loss": 0.0,
+      "step": 35570
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.1171702284450995e-08,
+      "loss": 0.0001,
+      "step": 35580
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.1024318349299926e-08,
+      "loss": 0.0,
+      "step": 35590
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.0876934414148856e-08,
+      "loss": 0.0,
+      "step": 35600
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.072955047899779e-08,
+      "loss": 0.0,
+      "step": 35610
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.058216654384672e-08,
+      "loss": 0.0,
+      "step": 35620
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.0434782608695655e-08,
+      "loss": 0.0,
+      "step": 35630
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.0287398673544585e-08,
+      "loss": 0.0001,
+      "step": 35640
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 3.0140014738393516e-08,
+      "loss": 0.0001,
+      "step": 35650
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 2.9992630803242446e-08,
+      "loss": 0.0001,
+      "step": 35660
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 2.984524686809138e-08,
+      "loss": 0.0,
+      "step": 35670
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 2.969786293294031e-08,
+      "loss": 0.0001,
+      "step": 35680
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 2.955047899778924e-08,
+      "loss": 0.0021,
+      "step": 35690
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 2.940309506263817e-08,
+      "loss": 0.0001,
+      "step": 35700
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 2.9255711127487102e-08,
+      "loss": 0.0001,
+      "step": 35710
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 2.9108327192336036e-08,
+      "loss": 0.0,
+      "step": 35720
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 2.8960943257184967e-08,
+      "loss": 0.0,
+      "step": 35730
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 2.8813559322033897e-08,
+      "loss": 0.0001,
+      "step": 35740
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 2.8666175386882828e-08,
+      "loss": 0.0001,
+      "step": 35750
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 2.851879145173176e-08,
+      "loss": 0.0001,
+      "step": 35760
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.8371407516580692e-08,
+      "loss": 0.0002,
+      "step": 35770
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.8224023581429623e-08,
+      "loss": 0.1711,
+      "step": 35780
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.8076639646278553e-08,
+      "loss": 0.0001,
+      "step": 35790
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.7929255711127487e-08,
+      "loss": 0.1409,
+      "step": 35800
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.7781871775976418e-08,
+      "loss": 0.0,
+      "step": 35810
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.7634487840825348e-08,
+      "loss": 0.0001,
+      "step": 35820
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.748710390567428e-08,
+      "loss": 0.0,
+      "step": 35830
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.7339719970523212e-08,
+      "loss": 0.0003,
+      "step": 35840
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.7192336035372143e-08,
+      "loss": 0.0001,
+      "step": 35850
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.7044952100221074e-08,
+      "loss": 0.0,
+      "step": 35860
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.6897568165070004e-08,
+      "loss": 0.1603,
+      "step": 35870
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.6750184229918938e-08,
+      "loss": 0.0002,
+      "step": 35880
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.660280029476787e-08,
+      "loss": 0.0,
+      "step": 35890
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.64554163596168e-08,
+      "loss": 0.0,
+      "step": 35900
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.630803242446573e-08,
+      "loss": 0.2063,
+      "step": 35910
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.6160648489314663e-08,
+      "loss": 0.0001,
+      "step": 35920
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.6013264554163594e-08,
+      "loss": 0.1632,
+      "step": 35930
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.5865880619012524e-08,
+      "loss": 0.0,
+      "step": 35940
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.5718496683861455e-08,
+      "loss": 0.0,
+      "step": 35950
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.557111274871039e-08,
+      "loss": 0.0,
+      "step": 35960
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.5423728813559323e-08,
+      "loss": 0.0,
+      "step": 35970
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.5276344878408253e-08,
+      "loss": 0.0001,
+      "step": 35980
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.5128960943257184e-08,
+      "loss": 0.0002,
+      "step": 35990
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.4981577008106118e-08,
+      "loss": 0.0,
+      "step": 36000
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.4834193072955048e-08,
+      "loss": 0.0001,
+      "step": 36010
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.468680913780398e-08,
+      "loss": 0.0,
+      "step": 36020
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.453942520265291e-08,
+      "loss": 0.0001,
+      "step": 36030
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.4392041267501843e-08,
+      "loss": 0.1962,
+      "step": 36040
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.4244657332350774e-08,
+      "loss": 0.0001,
+      "step": 36050
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.4097273397199704e-08,
+      "loss": 0.0,
+      "step": 36060
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.3949889462048635e-08,
+      "loss": 0.0,
+      "step": 36070
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.380250552689757e-08,
+      "loss": 0.0,
+      "step": 36080
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.36551215917465e-08,
+      "loss": 0.0,
+      "step": 36090
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.350773765659543e-08,
+      "loss": 0.0001,
+      "step": 36100
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.336035372144436e-08,
+      "loss": 0.0027,
+      "step": 36110
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.3212969786293294e-08,
+      "loss": 0.1708,
+      "step": 36120
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 2.3065585851142225e-08,
+      "loss": 0.0,
+      "step": 36130
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 2.2918201915991155e-08,
+      "loss": 0.0002,
+      "step": 36140
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 2.277081798084009e-08,
+      "loss": 0.0,
+      "step": 36150
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 2.262343404568902e-08,
+      "loss": 0.1789,
+      "step": 36160
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 2.247605011053795e-08,
+      "loss": 0.0,
+      "step": 36170
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 2.232866617538688e-08,
+      "loss": 0.0,
+      "step": 36180
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 2.2181282240235815e-08,
+      "loss": 0.0,
+      "step": 36190
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 2.2033898305084745e-08,
+      "loss": 0.0,
+      "step": 36200
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 2.1886514369933676e-08,
+      "loss": 0.0459,
+      "step": 36210
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 2.1739130434782606e-08,
+      "loss": 0.1254,
+      "step": 36220
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 2.159174649963154e-08,
+      "loss": 0.0,
+      "step": 36230
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 2.144436256448047e-08,
+      "loss": 0.0,
+      "step": 36240
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 2.12969786293294e-08,
+      "loss": 0.0,
+      "step": 36250
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 2.114959469417833e-08,
+      "loss": 0.0,
+      "step": 36260
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 2.1002210759027265e-08,
+      "loss": 0.0,
+      "step": 36270
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 2.0854826823876196e-08,
+      "loss": 0.0001,
+      "step": 36280
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 2.0707442888725127e-08,
+      "loss": 0.0,
+      "step": 36290
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 2.0560058953574057e-08,
+      "loss": 0.0001,
+      "step": 36300
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 2.041267501842299e-08,
+      "loss": 0.0001,
+      "step": 36310
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 2.026529108327192e-08,
+      "loss": 0.0059,
+      "step": 36320
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 2.0117907148120852e-08,
+      "loss": 0.0001,
+      "step": 36330
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.9970523212969783e-08,
+      "loss": 0.0,
+      "step": 36340
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.982313927781872e-08,
+      "loss": 0.0,
+      "step": 36350
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.967575534266765e-08,
+      "loss": 0.0,
+      "step": 36360
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.952837140751658e-08,
+      "loss": 0.1296,
+      "step": 36370
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.938098747236551e-08,
+      "loss": 0.0001,
+      "step": 36380
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.9233603537214445e-08,
+      "loss": 0.0,
+      "step": 36390
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.9086219602063376e-08,
+      "loss": 0.0001,
+      "step": 36400
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.8938835666912306e-08,
+      "loss": 0.0,
+      "step": 36410
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.8791451731761237e-08,
+      "loss": 0.0001,
+      "step": 36420
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.864406779661017e-08,
+      "loss": 0.0,
+      "step": 36430
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.84966838614591e-08,
+      "loss": 0.0002,
+      "step": 36440
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.8349299926308032e-08,
+      "loss": 0.0001,
+      "step": 36450
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.8201915991156962e-08,
+      "loss": 0.1939,
+      "step": 36460
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.8054532056005896e-08,
+      "loss": 0.1656,
+      "step": 36470
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.7907148120854827e-08,
+      "loss": 0.0,
+      "step": 36480
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.7759764185703757e-08,
+      "loss": 0.0,
+      "step": 36490
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.7612380250552688e-08,
+      "loss": 0.0001,
+      "step": 36500
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.746499631540162e-08,
+      "loss": 0.1785,
+      "step": 36510
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.7317612380250552e-08,
+      "loss": 0.0,
+      "step": 36520
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.7170228445099483e-08,
+      "loss": 0.0001,
+      "step": 36530
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.7022844509948413e-08,
+      "loss": 0.0,
+      "step": 36540
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.6875460574797347e-08,
+      "loss": 0.0005,
+      "step": 36550
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.6728076639646278e-08,
+      "loss": 0.0,
+      "step": 36560
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.6580692704495208e-08,
+      "loss": 0.0,
+      "step": 36570
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.643330876934414e-08,
+      "loss": 0.0001,
+      "step": 36580
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.6285924834193073e-08,
+      "loss": 0.0001,
+      "step": 36590
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.6138540899042003e-08,
+      "loss": 0.0001,
+      "step": 36600
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.5991156963890934e-08,
+      "loss": 0.0001,
+      "step": 36610
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.5843773028739868e-08,
+      "loss": 0.0006,
+      "step": 36620
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.5696389093588798e-08,
+      "loss": 0.0,
+      "step": 36630
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.5549005158437732e-08,
+      "loss": 0.2692,
+      "step": 36640
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.5401621223286662e-08,
+      "loss": 0.0,
+      "step": 36650
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.5254237288135593e-08,
+      "loss": 0.0,
+      "step": 36660
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.5106853352984524e-08,
+      "loss": 0.1967,
+      "step": 36670
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.4959469417833457e-08,
+      "loss": 0.0001,
+      "step": 36680
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.4812085482682388e-08,
+      "loss": 0.0,
+      "step": 36690
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.4664701547531318e-08,
+      "loss": 0.0,
+      "step": 36700
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.451731761238025e-08,
+      "loss": 0.0,
+      "step": 36710
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.4369933677229181e-08,
+      "loss": 0.1343,
+      "step": 36720
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.4222549742078113e-08,
+      "loss": 0.0001,
+      "step": 36730
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.4075165806927044e-08,
+      "loss": 0.0001,
+      "step": 36740
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.3927781871775976e-08,
+      "loss": 0.0,
+      "step": 36750
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.3780397936624907e-08,
+      "loss": 0.0001,
+      "step": 36760
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.3633014001473839e-08,
+      "loss": 0.0,
+      "step": 36770
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.348563006632277e-08,
+      "loss": 0.0001,
+      "step": 36780
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.3338246131171702e-08,
+      "loss": 0.0,
+      "step": 36790
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.3190862196020632e-08,
+      "loss": 0.0,
+      "step": 36800
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.3043478260869564e-08,
+      "loss": 0.0001,
+      "step": 36810
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.2896094325718495e-08,
+      "loss": 0.0,
+      "step": 36820
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.2748710390567429e-08,
+      "loss": 0.2478,
+      "step": 36830
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.260132645541636e-08,
+      "loss": 0.0001,
+      "step": 36840
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.2453942520265291e-08,
+      "loss": 0.0033,
+      "step": 36850
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.2306558585114222e-08,
+      "loss": 0.0001,
+      "step": 36860
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.2159174649963154e-08,
+      "loss": 0.0,
+      "step": 36870
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.2011790714812085e-08,
+      "loss": 0.0004,
+      "step": 36880
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.1864406779661017e-08,
+      "loss": 0.0001,
+      "step": 36890
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 1.1717022844509947e-08,
+      "loss": 0.0002,
+      "step": 36900
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 1.156963890935888e-08,
+      "loss": 0.1942,
+      "step": 36910
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 1.142225497420781e-08,
+      "loss": 0.0,
+      "step": 36920
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 1.1274871039056742e-08,
+      "loss": 0.0002,
+      "step": 36930
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 1.1127487103905673e-08,
+      "loss": 0.0,
+      "step": 36940
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 1.0980103168754605e-08,
+      "loss": 0.0001,
+      "step": 36950
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 1.0832719233603536e-08,
+      "loss": 0.0,
+      "step": 36960
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 1.0685335298452468e-08,
+      "loss": 0.0021,
+      "step": 36970
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 1.0537951363301398e-08,
+      "loss": 0.0001,
+      "step": 36980
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 1.039056742815033e-08,
+      "loss": 0.0,
+      "step": 36990
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 1.0243183492999261e-08,
+      "loss": 0.0001,
+      "step": 37000
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 1.0095799557848193e-08,
+      "loss": 0.0,
+      "step": 37010
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 9.948415622697124e-09,
+      "loss": 0.0001,
+      "step": 37020
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 9.801031687546058e-09,
+      "loss": 0.0,
+      "step": 37030
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 9.65364775239499e-09,
+      "loss": 0.0001,
+      "step": 37040
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 9.50626381724392e-09,
+      "loss": 0.0001,
+      "step": 37050
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 9.358879882092853e-09,
+      "loss": 0.0001,
+      "step": 37060
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 9.211495946941783e-09,
+      "loss": 0.0,
+      "step": 37070
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 9.064112011790715e-09,
+      "loss": 0.0001,
+      "step": 37080
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 8.916728076639646e-09,
+      "loss": 0.0,
+      "step": 37090
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 8.769344141488578e-09,
+      "loss": 0.1614,
+      "step": 37100
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 8.621960206337509e-09,
+      "loss": 0.0001,
+      "step": 37110
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 8.474576271186441e-09,
+      "loss": 0.0,
+      "step": 37120
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 8.327192336035371e-09,
+      "loss": 0.0,
+      "step": 37130
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 8.179808400884304e-09,
+      "loss": 0.0,
+      "step": 37140
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 8.032424465733234e-09,
+      "loss": 0.0001,
+      "step": 37150
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 7.885040530582166e-09,
+      "loss": 0.1901,
+      "step": 37160
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 7.737656595431099e-09,
+      "loss": 0.0,
+      "step": 37170
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 7.590272660280029e-09,
+      "loss": 0.185,
+      "step": 37180
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 7.4428887251289605e-09,
+      "loss": 0.0001,
+      "step": 37190
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 7.295504789977892e-09,
+      "loss": 0.0001,
+      "step": 37200
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 7.148120854826823e-09,
+      "loss": 0.0,
+      "step": 37210
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 7.000736919675755e-09,
+      "loss": 0.0001,
+      "step": 37220
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 6.853352984524686e-09,
+      "loss": 0.0001,
+      "step": 37230
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 6.705969049373617e-09,
+      "loss": 0.1437,
+      "step": 37240
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 6.558585114222549e-09,
+      "loss": 0.0,
+      "step": 37250
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 6.411201179071481e-09,
+      "loss": 0.0,
+      "step": 37260
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 6.263817243920413e-09,
+      "loss": 0.0,
+      "step": 37270
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 6.1164333087693445e-09,
+      "loss": 0.1345,
+      "step": 37280
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 5.969049373618276e-09,
+      "loss": 0.0021,
+      "step": 37290
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 5.821665438467207e-09,
+      "loss": 0.0001,
+      "step": 37300
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 5.6742815033161386e-09,
+      "loss": 0.0014,
+      "step": 37310
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 5.52689756816507e-09,
+      "loss": 0.0,
+      "step": 37320
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 5.379513633014001e-09,
+      "loss": 0.0001,
+      "step": 37330
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 5.232129697862933e-09,
+      "loss": 0.1849,
+      "step": 37340
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 5.084745762711864e-09,
+      "loss": 0.1349,
+      "step": 37350
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 4.937361827560796e-09,
+      "loss": 0.0001,
+      "step": 37360
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 4.789977892409728e-09,
+      "loss": 0.0,
+      "step": 37370
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 4.642593957258659e-09,
+      "loss": 0.0001,
+      "step": 37380
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 4.49521002210759e-09,
+      "loss": 0.0,
+      "step": 37390
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 4.347826086956522e-09,
+      "loss": 0.1958,
+      "step": 37400
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 4.200442151805453e-09,
+      "loss": 0.1018,
+      "step": 37410
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 4.0530582166543845e-09,
+      "loss": 0.0,
+      "step": 37420
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.905674281503316e-09,
+      "loss": 0.0001,
+      "step": 37430
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.758290346352247e-09,
+      "loss": 0.0,
+      "step": 37440
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.610906411201179e-09,
+      "loss": 0.0001,
+      "step": 37450
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.4635224760501103e-09,
+      "loss": 0.1866,
+      "step": 37460
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.3161385408990417e-09,
+      "loss": 0.0001,
+      "step": 37470
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.1687546057479735e-09,
+      "loss": 0.0,
+      "step": 37480
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 3.021370670596905e-09,
+      "loss": 0.0,
+      "step": 37490
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 2.8739867354458362e-09,
+      "loss": 0.0,
+      "step": 37500
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 2.7266028002947676e-09,
+      "loss": 0.0001,
+      "step": 37510
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 2.579218865143699e-09,
+      "loss": 0.0,
+      "step": 37520
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 2.4318349299926308e-09,
+      "loss": 0.0001,
+      "step": 37530
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 2.284450994841562e-09,
+      "loss": 0.0,
+      "step": 37540
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 2.1370670596904935e-09,
+      "loss": 0.0001,
+      "step": 37550
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 1.989683124539425e-09,
+      "loss": 0.0947,
+      "step": 37560
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 1.8422991893883566e-09,
+      "loss": 0.0,
+      "step": 37570
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 1.694915254237288e-09,
+      "loss": 0.0001,
+      "step": 37580
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 1.5475313190862196e-09,
+      "loss": 0.0036,
+      "step": 37590
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 1.400147383935151e-09,
+      "loss": 0.0001,
+      "step": 37600
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 1.2527634487840823e-09,
+      "loss": 0.0,
+      "step": 37610
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 1.105379513633014e-09,
+      "loss": 0.0,
+      "step": 37620
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 9.579955784819455e-10,
+      "loss": 0.0527,
+      "step": 37630
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 8.106116433308769e-10,
+      "loss": 0.0,
+      "step": 37640
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 6.632277081798083e-10,
+      "loss": 0.0001,
+      "step": 37650
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 5.158437730287398e-10,
+      "loss": 0.0,
+      "step": 37660
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 3.684598378776713e-10,
+      "loss": 0.0,
+      "step": 37670
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 2.210759027266028e-10,
+      "loss": 0.0001,
+      "step": 37680
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 7.369196757553426e-11,
+      "loss": 0.0001,
+      "step": 37690
+    },
+    {
+      "epoch": 14.07,
+      "eval_accuracy": 0.8984711877695022,
+      "eval_loss": 0.8329455852508545,
+      "eval_runtime": 592.8076,
+      "eval_samples_per_second": 4.303,
+      "eval_steps_per_second": 1.076,
+      "step": 37695
+    },
+    {
+      "epoch": 14.07,
+      "step": 37695,
+      "total_flos": 6.620174734520327e+20,
+      "train_loss": 0.16967469725597836,
+      "train_runtime": 72066.3663,
+      "train_samples_per_second": 2.092,
+      "train_steps_per_second": 0.523
+    },
+    {
+      "epoch": 14.07,
+      "eval_accuracy": 0.8992551940415523,
+      "eval_loss": 0.7024816274642944,
+      "eval_runtime": 675.1744,
+      "eval_samples_per_second": 3.778,
+      "eval_steps_per_second": 0.945,
+      "step": 37695
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 37695,
+  "num_train_epochs": 9223372036854775807,
+  "save_steps": 500,
+  "total_flos": 6.620174734520327e+20,
+  "trial_name": null,
+  "trial_params": null
+}