diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,48745 @@
+{
+  "best_metric": 0.0124629195779562,
+  "best_model_checkpoint": "./save/eng-zho_others_facebook/wav2vec2-large-xlsr-53/checkpoint-27000",
+  "epoch": 15.0,
+  "global_step": 81000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999907407407408e-05,
+      "loss": 0.0531,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999824074074074e-05,
+      "loss": 0.0518,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999731481481482e-05,
+      "loss": 0.0507,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999648148148148e-05,
+      "loss": 0.0485,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999555555555556e-05,
+      "loss": 0.0456,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999462962962963e-05,
+      "loss": 0.0421,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99937037037037e-05,
+      "loss": 0.0391,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999277777777778e-05,
+      "loss": 0.0371,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999185185185186e-05,
+      "loss": 0.0369,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999092592592593e-05,
+      "loss": 0.0372,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999e-05,
+      "loss": 0.0359,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9989074074074076e-05,
+      "loss": 0.0371,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998814814814815e-05,
+      "loss": 0.0354,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9987222222222226e-05,
+      "loss": 0.0375,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99862962962963e-05,
+      "loss": 0.0357,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998537037037037e-05,
+      "loss": 0.0363,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9984444444444444e-05,
+      "loss": 0.0359,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998351851851852e-05,
+      "loss": 0.037,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9982592592592594e-05,
+      "loss": 0.0348,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998166666666667e-05,
+      "loss": 0.0348,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9980740740740745e-05,
+      "loss": 0.0358,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997981481481482e-05,
+      "loss": 0.0363,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9978888888888895e-05,
+      "loss": 0.035,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997796296296296e-05,
+      "loss": 0.0361,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.997703703703704e-05,
+      "loss": 0.0344,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.997611111111111e-05,
+      "loss": 0.0353,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.997518518518519e-05,
+      "loss": 0.0355,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9974259259259256e-05,
+      "loss": 0.0344,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.997333333333333e-05,
+      "loss": 0.0354,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.997240740740741e-05,
+      "loss": 0.0348,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.997148148148149e-05,
+      "loss": 0.0346,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.997055555555556e-05,
+      "loss": 0.0336,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.996962962962963e-05,
+      "loss": 0.0329,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.996870370370371e-05,
+      "loss": 0.0334,
+      "step": 340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.996777777777778e-05,
+      "loss": 0.0338,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.996685185185186e-05,
+      "loss": 0.0326,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9965925925925925e-05,
+      "loss": 0.0311,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9965e-05,
+      "loss": 0.0338,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9964074074074075e-05,
+      "loss": 0.034,
+      "step": 390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.996314814814815e-05,
+      "loss": 0.0337,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9962222222222225e-05,
+      "loss": 0.033,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.99612962962963e-05,
+      "loss": 0.0318,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9960370370370375e-05,
+      "loss": 0.0328,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.995944444444445e-05,
+      "loss": 0.0309,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.995851851851852e-05,
+      "loss": 0.0319,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9957592592592594e-05,
+      "loss": 0.031,
+      "step": 460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.995666666666667e-05,
+      "loss": 0.0313,
+      "step": 470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9955740740740744e-05,
+      "loss": 0.0317,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.995481481481481e-05,
+      "loss": 0.0332,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.995388888888889e-05,
+      "loss": 0.0304,
+      "step": 500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.995296296296297e-05,
+      "loss": 0.0302,
+      "step": 510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9952037037037044e-05,
+      "loss": 0.0314,
+      "step": 520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.995111111111111e-05,
+      "loss": 0.0304,
+      "step": 530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.995018518518519e-05,
+      "loss": 0.0314,
+      "step": 540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.994925925925926e-05,
+      "loss": 0.0333,
+      "step": 550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.994833333333334e-05,
+      "loss": 0.0313,
+      "step": 560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.994740740740741e-05,
+      "loss": 0.031,
+      "step": 570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.994648148148148e-05,
+      "loss": 0.0317,
+      "step": 580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9945555555555556e-05,
+      "loss": 0.0304,
+      "step": 590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.994462962962963e-05,
+      "loss": 0.0313,
+      "step": 600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9943703703703706e-05,
+      "loss": 0.0278,
+      "step": 610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.994277777777778e-05,
+      "loss": 0.0285,
+      "step": 620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9941851851851856e-05,
+      "loss": 0.0296,
+      "step": 630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.994092592592593e-05,
+      "loss": 0.0302,
+      "step": 640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9940000000000006e-05,
+      "loss": 0.0317,
+      "step": 650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9939074074074075e-05,
+      "loss": 0.0291,
+      "step": 660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.993814814814815e-05,
+      "loss": 0.0306,
+      "step": 670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9937222222222225e-05,
+      "loss": 0.0302,
+      "step": 680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.99362962962963e-05,
+      "loss": 0.0278,
+      "step": 690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.993537037037037e-05,
+      "loss": 0.03,
+      "step": 700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.993444444444444e-05,
+      "loss": 0.0291,
+      "step": 710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.993351851851852e-05,
+      "loss": 0.0286,
+      "step": 720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.99325925925926e-05,
+      "loss": 0.0297,
+      "step": 730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.993166666666667e-05,
+      "loss": 0.0286,
+      "step": 740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.993074074074074e-05,
+      "loss": 0.0299,
+      "step": 750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.992981481481482e-05,
+      "loss": 0.0284,
+      "step": 760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9928888888888893e-05,
+      "loss": 0.0289,
+      "step": 770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.992796296296297e-05,
+      "loss": 0.0284,
+      "step": 780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.992703703703704e-05,
+      "loss": 0.0279,
+      "step": 790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.992611111111111e-05,
+      "loss": 0.0295,
+      "step": 800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.992518518518519e-05,
+      "loss": 0.0285,
+      "step": 810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.992425925925926e-05,
+      "loss": 0.0258,
+      "step": 820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.992333333333333e-05,
+      "loss": 0.0273,
+      "step": 830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.992240740740741e-05,
+      "loss": 0.0274,
+      "step": 840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.992148148148149e-05,
+      "loss": 0.0248,
+      "step": 850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.992055555555556e-05,
+      "loss": 0.0285,
+      "step": 860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.991962962962963e-05,
+      "loss": 0.0265,
+      "step": 870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9918703703703706e-05,
+      "loss": 0.0271,
+      "step": 880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.991777777777778e-05,
+      "loss": 0.0289,
+      "step": 890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9916851851851856e-05,
+      "loss": 0.0268,
+      "step": 900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9915925925925924e-05,
+      "loss": 0.0271,
+      "step": 910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9915e-05,
+      "loss": 0.0274,
+      "step": 920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9914074074074074e-05,
+      "loss": 0.0254,
+      "step": 930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.991314814814815e-05,
+      "loss": 0.0265,
+      "step": 940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9912222222222224e-05,
+      "loss": 0.0246,
+      "step": 950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.99112962962963e-05,
+      "loss": 0.025,
+      "step": 960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9910370370370374e-05,
+      "loss": 0.0259,
+      "step": 970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.990944444444445e-05,
+      "loss": 0.0263,
+      "step": 980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9908518518518524e-05,
+      "loss": 0.0228,
+      "step": 990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.990759259259259e-05,
+      "loss": 0.0235,
+      "step": 1000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.990666666666667e-05,
+      "loss": 0.0245,
+      "step": 1010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.990574074074074e-05,
+      "loss": 0.0278,
+      "step": 1020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.990481481481482e-05,
+      "loss": 0.0244,
+      "step": 1030
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9903888888888886e-05,
+      "loss": 0.0259,
+      "step": 1040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.990296296296297e-05,
+      "loss": 0.0226,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.990203703703704e-05,
+      "loss": 0.0236,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.990111111111112e-05,
+      "loss": 0.0229,
+      "step": 1070
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9900185185185186e-05,
+      "loss": 0.0239,
+      "step": 1080
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.989925925925926e-05,
+      "loss": 0.0247,
+      "step": 1090
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9898333333333336e-05,
+      "loss": 0.0216,
+      "step": 1100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.989740740740741e-05,
+      "loss": 0.0222,
+      "step": 1110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.989648148148148e-05,
+      "loss": 0.021,
+      "step": 1120
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9895555555555555e-05,
+      "loss": 0.0219,
+      "step": 1130
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.989462962962963e-05,
+      "loss": 0.0243,
+      "step": 1140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9893703703703705e-05,
+      "loss": 0.0248,
+      "step": 1150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.989277777777778e-05,
+      "loss": 0.0202,
+      "step": 1160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9891851851851855e-05,
+      "loss": 0.0241,
+      "step": 1170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.989092592592593e-05,
+      "loss": 0.0252,
+      "step": 1180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9890000000000005e-05,
+      "loss": 0.025,
+      "step": 1190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.988907407407408e-05,
+      "loss": 0.0217,
+      "step": 1200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.988814814814815e-05,
+      "loss": 0.0248,
+      "step": 1210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9887222222222224e-05,
+      "loss": 0.0224,
+      "step": 1220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.98862962962963e-05,
+      "loss": 0.0219,
+      "step": 1230
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9885370370370374e-05,
+      "loss": 0.0208,
+      "step": 1240
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.988444444444444e-05,
+      "loss": 0.021,
+      "step": 1250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.988351851851852e-05,
+      "loss": 0.0214,
+      "step": 1260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.98825925925926e-05,
+      "loss": 0.0237,
+      "step": 1270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9881666666666674e-05,
+      "loss": 0.0204,
+      "step": 1280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.988074074074074e-05,
+      "loss": 0.0236,
+      "step": 1290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.987981481481482e-05,
+      "loss": 0.0205,
+      "step": 1300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.987888888888889e-05,
+      "loss": 0.0219,
+      "step": 1310
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.987796296296297e-05,
+      "loss": 0.0198,
+      "step": 1320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9877037037037036e-05,
+      "loss": 0.0209,
+      "step": 1330
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.987611111111111e-05,
+      "loss": 0.0187,
+      "step": 1340
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9875185185185186e-05,
+      "loss": 0.0224,
+      "step": 1350
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.987425925925926e-05,
+      "loss": 0.0229,
+      "step": 1360
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9873333333333336e-05,
+      "loss": 0.0214,
+      "step": 1370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.987240740740741e-05,
+      "loss": 0.021,
+      "step": 1380
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9871481481481486e-05,
+      "loss": 0.0203,
+      "step": 1390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.987055555555556e-05,
+      "loss": 0.0203,
+      "step": 1400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9869629629629636e-05,
+      "loss": 0.0225,
+      "step": 1410
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9868703703703704e-05,
+      "loss": 0.0186,
+      "step": 1420
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9867870370370376e-05,
+      "loss": 0.0196,
+      "step": 1430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9866944444444444e-05,
+      "loss": 0.0203,
+      "step": 1440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.986601851851852e-05,
+      "loss": 0.0202,
+      "step": 1450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9865092592592594e-05,
+      "loss": 0.0201,
+      "step": 1460
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.986416666666667e-05,
+      "loss": 0.0214,
+      "step": 1470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9863333333333334e-05,
+      "loss": 0.022,
+      "step": 1480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.986240740740741e-05,
+      "loss": 0.0202,
+      "step": 1490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9861481481481484e-05,
+      "loss": 0.0216,
+      "step": 1500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.986055555555556e-05,
+      "loss": 0.0244,
+      "step": 1510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9859629629629634e-05,
+      "loss": 0.0215,
+      "step": 1520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.985870370370371e-05,
+      "loss": 0.021,
+      "step": 1530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.985777777777778e-05,
+      "loss": 0.0195,
+      "step": 1540
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.985685185185185e-05,
+      "loss": 0.0217,
+      "step": 1550
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.985592592592593e-05,
+      "loss": 0.0203,
+      "step": 1560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9855e-05,
+      "loss": 0.0207,
+      "step": 1570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.985407407407408e-05,
+      "loss": 0.0204,
+      "step": 1580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9853148148148146e-05,
+      "loss": 0.0193,
+      "step": 1590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.985222222222222e-05,
+      "loss": 0.0183,
+      "step": 1600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9851296296296296e-05,
+      "loss": 0.0174,
+      "step": 1610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.985037037037038e-05,
+      "loss": 0.0246,
+      "step": 1620
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9849444444444446e-05,
+      "loss": 0.0198,
+      "step": 1630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.984851851851852e-05,
+      "loss": 0.0204,
+      "step": 1640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9847592592592597e-05,
+      "loss": 0.0198,
+      "step": 1650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.984666666666667e-05,
+      "loss": 0.0222,
+      "step": 1660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.984574074074074e-05,
+      "loss": 0.0191,
+      "step": 1670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9844814814814815e-05,
+      "loss": 0.0198,
+      "step": 1680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.984388888888889e-05,
+      "loss": 0.0182,
+      "step": 1690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9842962962962965e-05,
+      "loss": 0.0162,
+      "step": 1700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984203703703704e-05,
+      "loss": 0.0168,
+      "step": 1710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9841111111111115e-05,
+      "loss": 0.0181,
+      "step": 1720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984018518518519e-05,
+      "loss": 0.0215,
+      "step": 1730
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9839259259259265e-05,
+      "loss": 0.0194,
+      "step": 1740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9838333333333334e-05,
+      "loss": 0.0195,
+      "step": 1750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983740740740741e-05,
+      "loss": 0.0173,
+      "step": 1760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9836481481481484e-05,
+      "loss": 0.02,
+      "step": 1770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983555555555556e-05,
+      "loss": 0.0215,
+      "step": 1780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9834629629629634e-05,
+      "loss": 0.0154,
+      "step": 1790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.98337037037037e-05,
+      "loss": 0.0211,
+      "step": 1800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.983277777777778e-05,
+      "loss": 0.0178,
+      "step": 1810
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.983185185185185e-05,
+      "loss": 0.0172,
+      "step": 1820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9830925925925934e-05,
+      "loss": 0.0169,
+      "step": 1830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.983e-05,
+      "loss": 0.0173,
+      "step": 1840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.982907407407408e-05,
+      "loss": 0.0153,
+      "step": 1850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.982814814814815e-05,
+      "loss": 0.0198,
+      "step": 1860
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.982722222222223e-05,
+      "loss": 0.0202,
+      "step": 1870
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9826296296296296e-05,
+      "loss": 0.0146,
+      "step": 1880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.982537037037037e-05,
+      "loss": 0.0171,
+      "step": 1890
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9824444444444446e-05,
+      "loss": 0.0176,
+      "step": 1900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.982351851851852e-05,
+      "loss": 0.0182,
+      "step": 1910
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9822592592592596e-05,
+      "loss": 0.0158,
+      "step": 1920
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9821666666666664e-05,
+      "loss": 0.0198,
+      "step": 1930
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9820740740740746e-05,
+      "loss": 0.0172,
+      "step": 1940
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.981981481481482e-05,
+      "loss": 0.018,
+      "step": 1950
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.981888888888889e-05,
+      "loss": 0.0154,
+      "step": 1960
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9817962962962964e-05,
+      "loss": 0.0194,
+      "step": 1970
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.981703703703704e-05,
+      "loss": 0.0164,
+      "step": 1980
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9816111111111115e-05,
+      "loss": 0.0185,
+      "step": 1990
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.981518518518519e-05,
+      "loss": 0.0168,
+      "step": 2000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.981425925925926e-05,
+      "loss": 0.0189,
+      "step": 2010
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.981333333333333e-05,
+      "loss": 0.0176,
+      "step": 2020
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.981240740740741e-05,
+      "loss": 0.0178,
+      "step": 2030
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.981148148148148e-05,
+      "loss": 0.0209,
+      "step": 2040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.981055555555556e-05,
+      "loss": 0.0157,
+      "step": 2050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.980962962962963e-05,
+      "loss": 0.021,
+      "step": 2060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.980870370370371e-05,
+      "loss": 0.0175,
+      "step": 2070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.980777777777778e-05,
+      "loss": 0.0169,
+      "step": 2080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.980685185185185e-05,
+      "loss": 0.0188,
+      "step": 2090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.980592592592593e-05,
+      "loss": 0.016,
+      "step": 2100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9805e-05,
+      "loss": 0.0168,
+      "step": 2110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.980407407407408e-05,
+      "loss": 0.0145,
+      "step": 2120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.980314814814815e-05,
+      "loss": 0.0149,
+      "step": 2130
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.980222222222222e-05,
+      "loss": 0.0156,
+      "step": 2140
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.98012962962963e-05,
+      "loss": 0.0217,
+      "step": 2150
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.980037037037038e-05,
+      "loss": 0.0161,
+      "step": 2160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9799444444444445e-05,
+      "loss": 0.0191,
+      "step": 2170
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.979851851851852e-05,
+      "loss": 0.023,
+      "step": 2180
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9797592592592595e-05,
+      "loss": 0.0155,
+      "step": 2190
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.979666666666667e-05,
+      "loss": 0.0181,
+      "step": 2200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9795740740740746e-05,
+      "loss": 0.0162,
+      "step": 2210
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9794814814814814e-05,
+      "loss": 0.0159,
+      "step": 2220
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.979388888888889e-05,
+      "loss": 0.0158,
+      "step": 2230
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9792962962962964e-05,
+      "loss": 0.014,
+      "step": 2240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.979203703703704e-05,
+      "loss": 0.0138,
+      "step": 2250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9791111111111114e-05,
+      "loss": 0.0158,
+      "step": 2260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.979018518518519e-05,
+      "loss": 0.0149,
+      "step": 2270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9789259259259264e-05,
+      "loss": 0.0165,
+      "step": 2280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.978833333333334e-05,
+      "loss": 0.0178,
+      "step": 2290
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.978740740740741e-05,
+      "loss": 0.0189,
+      "step": 2300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.978648148148148e-05,
+      "loss": 0.0169,
+      "step": 2310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.978555555555556e-05,
+      "loss": 0.0157,
+      "step": 2320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.978462962962963e-05,
+      "loss": 0.0126,
+      "step": 2330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.978370370370371e-05,
+      "loss": 0.0144,
+      "step": 2340
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9782777777777776e-05,
+      "loss": 0.0127,
+      "step": 2350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.978185185185185e-05,
+      "loss": 0.0152,
+      "step": 2360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.978092592592593e-05,
+      "loss": 0.0144,
+      "step": 2370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.978e-05,
+      "loss": 0.018,
+      "step": 2380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9779074074074076e-05,
+      "loss": 0.0145,
+      "step": 2390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.977814814814815e-05,
+      "loss": 0.0164,
+      "step": 2400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9777222222222226e-05,
+      "loss": 0.0153,
+      "step": 2410
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.97762962962963e-05,
+      "loss": 0.015,
+      "step": 2420
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.977537037037037e-05,
+      "loss": 0.0183,
+      "step": 2430
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9774444444444445e-05,
+      "loss": 0.0172,
+      "step": 2440
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.977351851851852e-05,
+      "loss": 0.0173,
+      "step": 2450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9772592592592595e-05,
+      "loss": 0.0159,
+      "step": 2460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.977166666666666e-05,
+      "loss": 0.0146,
+      "step": 2470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9770740740740745e-05,
+      "loss": 0.0172,
+      "step": 2480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.976981481481482e-05,
+      "loss": 0.0133,
+      "step": 2490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9768888888888895e-05,
+      "loss": 0.0203,
+      "step": 2500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.976796296296296e-05,
+      "loss": 0.0114,
+      "step": 2510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.976703703703704e-05,
+      "loss": 0.0154,
+      "step": 2520
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9766111111111113e-05,
+      "loss": 0.0155,
+      "step": 2530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.976518518518519e-05,
+      "loss": 0.0192,
+      "step": 2540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9764259259259264e-05,
+      "loss": 0.0145,
+      "step": 2550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.976333333333333e-05,
+      "loss": 0.0127,
+      "step": 2560
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.976240740740741e-05,
+      "loss": 0.0182,
+      "step": 2570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.976148148148148e-05,
+      "loss": 0.0148,
+      "step": 2580
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.976055555555556e-05,
+      "loss": 0.0156,
+      "step": 2590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.975962962962963e-05,
+      "loss": 0.0163,
+      "step": 2600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.975870370370371e-05,
+      "loss": 0.0142,
+      "step": 2610
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.975777777777778e-05,
+      "loss": 0.0151,
+      "step": 2620
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.975685185185186e-05,
+      "loss": 0.0191,
+      "step": 2630
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9755925925925925e-05,
+      "loss": 0.015,
+      "step": 2640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9755e-05,
+      "loss": 0.0164,
+      "step": 2650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9754074074074076e-05,
+      "loss": 0.0158,
+      "step": 2660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.975314814814815e-05,
+      "loss": 0.0159,
+      "step": 2670
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9752222222222226e-05,
+      "loss": 0.0176,
+      "step": 2680
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.97512962962963e-05,
+      "loss": 0.0181,
+      "step": 2690
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9750370370370376e-05,
+      "loss": 0.0149,
+      "step": 2700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.974944444444445e-05,
+      "loss": 0.0139,
+      "step": 2710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.974851851851852e-05,
+      "loss": 0.0133,
+      "step": 2720
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9747592592592594e-05,
+      "loss": 0.0126,
+      "step": 2730
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.974666666666667e-05,
+      "loss": 0.0131,
+      "step": 2740
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9745740740740744e-05,
+      "loss": 0.0161,
+      "step": 2750
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.974481481481482e-05,
+      "loss": 0.0172,
+      "step": 2760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.974388888888889e-05,
+      "loss": 0.0142,
+      "step": 2770
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.974296296296296e-05,
+      "loss": 0.0121,
+      "step": 2780
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.974203703703704e-05,
+      "loss": 0.0165,
+      "step": 2790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.974111111111111e-05,
+      "loss": 0.0164,
+      "step": 2800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.974018518518519e-05,
+      "loss": 0.0141,
+      "step": 2810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.973925925925926e-05,
+      "loss": 0.0117,
+      "step": 2820
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.973833333333334e-05,
+      "loss": 0.013,
+      "step": 2830
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.973740740740741e-05,
+      "loss": 0.0184,
+      "step": 2840
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.973648148148148e-05,
+      "loss": 0.0144,
+      "step": 2850
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9735555555555556e-05,
+      "loss": 0.0172,
+      "step": 2860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.973462962962963e-05,
+      "loss": 0.0139,
+      "step": 2870
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9733703703703706e-05,
+      "loss": 0.0158,
+      "step": 2880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.973277777777778e-05,
+      "loss": 0.0138,
+      "step": 2890
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.973185185185185e-05,
+      "loss": 0.0166,
+      "step": 2900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.973092592592593e-05,
+      "loss": 0.0124,
+      "step": 2910
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.973000000000001e-05,
+      "loss": 0.0132,
+      "step": 2920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9729074074074075e-05,
+      "loss": 0.0134,
+      "step": 2930
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.972814814814815e-05,
+      "loss": 0.0205,
+      "step": 2940
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9727222222222225e-05,
+      "loss": 0.0162,
+      "step": 2950
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.97262962962963e-05,
+      "loss": 0.0148,
+      "step": 2960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9725370370370375e-05,
+      "loss": 0.0147,
+      "step": 2970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9724444444444443e-05,
+      "loss": 0.013,
+      "step": 2980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.972351851851852e-05,
+      "loss": 0.0133,
+      "step": 2990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9722592592592594e-05,
+      "loss": 0.0155,
+      "step": 3000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.972166666666667e-05,
+      "loss": 0.0099,
+      "step": 3010
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9720740740740744e-05,
+      "loss": 0.0133,
+      "step": 3020
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.971981481481482e-05,
+      "loss": 0.0141,
+      "step": 3030
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9718888888888894e-05,
+      "loss": 0.013,
+      "step": 3040
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.971796296296297e-05,
+      "loss": 0.0126,
+      "step": 3050
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.971703703703704e-05,
+      "loss": 0.0147,
+      "step": 3060
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.971611111111111e-05,
+      "loss": 0.0131,
+      "step": 3070
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.971518518518519e-05,
+      "loss": 0.0158,
+      "step": 3080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.971425925925926e-05,
+      "loss": 0.0112,
+      "step": 3090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.971333333333334e-05,
+      "loss": 0.0114,
+      "step": 3100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9712407407407406e-05,
+      "loss": 0.0121,
+      "step": 3110
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.971148148148148e-05,
+      "loss": 0.015,
+      "step": 3120
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.971055555555556e-05,
+      "loss": 0.0165,
+      "step": 3130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.970962962962963e-05,
+      "loss": 0.0123,
+      "step": 3140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9708703703703706e-05,
+      "loss": 0.0133,
+      "step": 3150
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.970777777777778e-05,
+      "loss": 0.0149,
+      "step": 3160
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9706851851851856e-05,
+      "loss": 0.0157,
+      "step": 3170
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.970592592592593e-05,
+      "loss": 0.014,
+      "step": 3180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9705e-05,
+      "loss": 0.0125,
+      "step": 3190
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9704074074074074e-05,
+      "loss": 0.0131,
+      "step": 3200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.970314814814815e-05,
+      "loss": 0.0143,
+      "step": 3210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.9702222222222224e-05,
+      "loss": 0.0149,
+      "step": 3220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.97012962962963e-05,
+      "loss": 0.0126,
+      "step": 3230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.9700370370370375e-05,
+      "loss": 0.0171,
+      "step": 3240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.969944444444445e-05,
+      "loss": 0.012,
+      "step": 3250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.9698518518518525e-05,
+      "loss": 0.0094,
+      "step": 3260
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.969759259259259e-05,
+      "loss": 0.0141,
+      "step": 3270
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.969666666666667e-05,
+      "loss": 0.0142,
+      "step": 3280
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.969574074074074e-05,
+      "loss": 0.0128,
+      "step": 3290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.969481481481482e-05,
+      "loss": 0.0108,
+      "step": 3300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.969388888888889e-05,
+      "loss": 0.0146,
+      "step": 3310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.969296296296296e-05,
+      "loss": 0.0185,
+      "step": 3320
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.9692037037037037e-05,
+      "loss": 0.0146,
+      "step": 3330
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.969111111111112e-05,
+      "loss": 0.0123,
+      "step": 3340
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.969018518518519e-05,
+      "loss": 0.015,
+      "step": 3350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.968925925925926e-05,
+      "loss": 0.0109,
+      "step": 3360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.968833333333334e-05,
+      "loss": 0.013,
+      "step": 3370
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.968740740740741e-05,
+      "loss": 0.0118,
+      "step": 3380
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.968648148148149e-05,
+      "loss": 0.0141,
+      "step": 3390
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.9685555555555555e-05,
+      "loss": 0.0158,
+      "step": 3400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.968462962962963e-05,
+      "loss": 0.0112,
+      "step": 3410
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.9683703703703705e-05,
+      "loss": 0.0121,
+      "step": 3420
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.968277777777778e-05,
+      "loss": 0.0125,
+      "step": 3430
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.968185185185185e-05,
+      "loss": 0.0108,
+      "step": 3440
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.968092592592593e-05,
+      "loss": 0.0132,
+      "step": 3450
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.9680000000000005e-05,
+      "loss": 0.0139,
+      "step": 3460
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.967907407407408e-05,
+      "loss": 0.0126,
+      "step": 3470
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.967814814814815e-05,
+      "loss": 0.0141,
+      "step": 3480
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.9677222222222224e-05,
+      "loss": 0.0151,
+      "step": 3490
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.96762962962963e-05,
+      "loss": 0.0102,
+      "step": 3500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.9675370370370374e-05,
+      "loss": 0.0137,
+      "step": 3510
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.967444444444445e-05,
+      "loss": 0.0119,
+      "step": 3520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.967351851851852e-05,
+      "loss": 0.013,
+      "step": 3530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.967259259259259e-05,
+      "loss": 0.0124,
+      "step": 3540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.967166666666667e-05,
+      "loss": 0.0144,
+      "step": 3550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.967074074074074e-05,
+      "loss": 0.0147,
+      "step": 3560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.966981481481482e-05,
+      "loss": 0.0131,
+      "step": 3570
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.966888888888889e-05,
+      "loss": 0.0109,
+      "step": 3580
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.966796296296297e-05,
+      "loss": 0.0136,
+      "step": 3590
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.966703703703704e-05,
+      "loss": 0.0114,
+      "step": 3600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.966611111111111e-05,
+      "loss": 0.0125,
+      "step": 3610
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9665185185185186e-05,
+      "loss": 0.0151,
+      "step": 3620
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.966425925925926e-05,
+      "loss": 0.0105,
+      "step": 3630
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9663333333333336e-05,
+      "loss": 0.0091,
+      "step": 3640
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9662407407407404e-05,
+      "loss": 0.0118,
+      "step": 3650
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9661481481481486e-05,
+      "loss": 0.0154,
+      "step": 3660
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.966055555555556e-05,
+      "loss": 0.0161,
+      "step": 3670
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9659629629629636e-05,
+      "loss": 0.0124,
+      "step": 3680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9658703703703705e-05,
+      "loss": 0.0133,
+      "step": 3690
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.965777777777778e-05,
+      "loss": 0.0113,
+      "step": 3700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9656851851851855e-05,
+      "loss": 0.0147,
+      "step": 3710
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.965592592592593e-05,
+      "loss": 0.0112,
+      "step": 3720
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9655000000000005e-05,
+      "loss": 0.0122,
+      "step": 3730
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.965407407407407e-05,
+      "loss": 0.0186,
+      "step": 3740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.965314814814815e-05,
+      "loss": 0.0174,
+      "step": 3750
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.965222222222222e-05,
+      "loss": 0.0144,
+      "step": 3760
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.96512962962963e-05,
+      "loss": 0.0107,
+      "step": 3770
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.965037037037037e-05,
+      "loss": 0.0142,
+      "step": 3780
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.964944444444445e-05,
+      "loss": 0.0103,
+      "step": 3790
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9648518518518523e-05,
+      "loss": 0.011,
+      "step": 3800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.96475925925926e-05,
+      "loss": 0.0123,
+      "step": 3810
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.964666666666667e-05,
+      "loss": 0.0138,
+      "step": 3820
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.964574074074074e-05,
+      "loss": 0.014,
+      "step": 3830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.964481481481482e-05,
+      "loss": 0.013,
+      "step": 3840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.964388888888889e-05,
+      "loss": 0.0153,
+      "step": 3850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.964296296296296e-05,
+      "loss": 0.015,
+      "step": 3860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.9642037037037035e-05,
+      "loss": 0.0114,
+      "step": 3870
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.964111111111112e-05,
+      "loss": 0.01,
+      "step": 3880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.964018518518519e-05,
+      "loss": 0.0149,
+      "step": 3890
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.963925925925926e-05,
+      "loss": 0.0107,
+      "step": 3900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.9638333333333336e-05,
+      "loss": 0.0134,
+      "step": 3910
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.963740740740741e-05,
+      "loss": 0.0138,
+      "step": 3920
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.9636481481481486e-05,
+      "loss": 0.0149,
+      "step": 3930
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.963555555555556e-05,
+      "loss": 0.0167,
+      "step": 3940
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.963462962962963e-05,
+      "loss": 0.0139,
+      "step": 3950
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.9633703703703704e-05,
+      "loss": 0.0134,
+      "step": 3960
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.963277777777778e-05,
+      "loss": 0.0155,
+      "step": 3970
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9631851851851854e-05,
+      "loss": 0.0132,
+      "step": 3980
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.963092592592593e-05,
+      "loss": 0.0098,
+      "step": 3990
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9630000000000004e-05,
+      "loss": 0.0126,
+      "step": 4000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.962907407407408e-05,
+      "loss": 0.0124,
+      "step": 4010
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.9628148148148154e-05,
+      "loss": 0.0129,
+      "step": 4020
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.962722222222222e-05,
+      "loss": 0.0129,
+      "step": 4030
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.96262962962963e-05,
+      "loss": 0.0148,
+      "step": 4040
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.962537037037037e-05,
+      "loss": 0.0129,
+      "step": 4050
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.962444444444445e-05,
+      "loss": 0.0127,
+      "step": 4060
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9623518518518516e-05,
+      "loss": 0.0096,
+      "step": 4070
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.962259259259259e-05,
+      "loss": 0.0126,
+      "step": 4080
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9621666666666666e-05,
+      "loss": 0.0121,
+      "step": 4090
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.962074074074075e-05,
+      "loss": 0.0121,
+      "step": 4100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9619814814814816e-05,
+      "loss": 0.0144,
+      "step": 4110
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.961888888888889e-05,
+      "loss": 0.0114,
+      "step": 4120
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9617962962962966e-05,
+      "loss": 0.013,
+      "step": 4130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.961703703703704e-05,
+      "loss": 0.0126,
+      "step": 4140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.9616111111111117e-05,
+      "loss": 0.0109,
+      "step": 4150
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.9615185185185185e-05,
+      "loss": 0.0142,
+      "step": 4160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.961425925925926e-05,
+      "loss": 0.0152,
+      "step": 4170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.9613333333333335e-05,
+      "loss": 0.0108,
+      "step": 4180
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.961240740740741e-05,
+      "loss": 0.0137,
+      "step": 4190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9611481481481485e-05,
+      "loss": 0.0147,
+      "step": 4200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.961055555555556e-05,
+      "loss": 0.0124,
+      "step": 4210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9609629629629635e-05,
+      "loss": 0.0122,
+      "step": 4220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.960870370370371e-05,
+      "loss": 0.015,
+      "step": 4230
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.960777777777778e-05,
+      "loss": 0.0145,
+      "step": 4240
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.9606851851851854e-05,
+      "loss": 0.0123,
+      "step": 4250
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.960592592592593e-05,
+      "loss": 0.0119,
+      "step": 4260
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.9605000000000004e-05,
+      "loss": 0.0098,
+      "step": 4270
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.960407407407407e-05,
+      "loss": 0.0132,
+      "step": 4280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.960314814814815e-05,
+      "loss": 0.0104,
+      "step": 4290
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.960222222222222e-05,
+      "loss": 0.0094,
+      "step": 4300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9601296296296304e-05,
+      "loss": 0.0131,
+      "step": 4310
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.960037037037037e-05,
+      "loss": 0.0098,
+      "step": 4320
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.959944444444445e-05,
+      "loss": 0.0078,
+      "step": 4330
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.959851851851852e-05,
+      "loss": 0.012,
+      "step": 4340
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.95975925925926e-05,
+      "loss": 0.0149,
+      "step": 4350
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.959666666666667e-05,
+      "loss": 0.0153,
+      "step": 4360
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.959574074074074e-05,
+      "loss": 0.0149,
+      "step": 4370
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.9594814814814816e-05,
+      "loss": 0.0158,
+      "step": 4380
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.959388888888889e-05,
+      "loss": 0.0143,
+      "step": 4390
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.9592962962962966e-05,
+      "loss": 0.0098,
+      "step": 4400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.9592037037037034e-05,
+      "loss": 0.0109,
+      "step": 4410
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.9591111111111116e-05,
+      "loss": 0.0129,
+      "step": 4420
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.959018518518519e-05,
+      "loss": 0.0149,
+      "step": 4430
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.9589259259259266e-05,
+      "loss": 0.0136,
+      "step": 4440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.9588333333333334e-05,
+      "loss": 0.0106,
+      "step": 4450
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.958740740740741e-05,
+      "loss": 0.0118,
+      "step": 4460
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.9586481481481484e-05,
+      "loss": 0.0119,
+      "step": 4470
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.958555555555556e-05,
+      "loss": 0.014,
+      "step": 4480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.958462962962963e-05,
+      "loss": 0.0114,
+      "step": 4490
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.95837037037037e-05,
+      "loss": 0.0136,
+      "step": 4500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.958277777777778e-05,
+      "loss": 0.011,
+      "step": 4510
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.958185185185185e-05,
+      "loss": 0.0156,
+      "step": 4520
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.958092592592593e-05,
+      "loss": 0.0124,
+      "step": 4530
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.958e-05,
+      "loss": 0.0143,
+      "step": 4540
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.957907407407408e-05,
+      "loss": 0.0139,
+      "step": 4550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.957814814814815e-05,
+      "loss": 0.011,
+      "step": 4560
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.957722222222223e-05,
+      "loss": 0.0099,
+      "step": 4570
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.9576296296296296e-05,
+      "loss": 0.0095,
+      "step": 4580
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.957537037037037e-05,
+      "loss": 0.012,
+      "step": 4590
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.9574444444444447e-05,
+      "loss": 0.0107,
+      "step": 4600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.957351851851852e-05,
+      "loss": 0.0088,
+      "step": 4610
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.957259259259259e-05,
+      "loss": 0.0134,
+      "step": 4620
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.957166666666667e-05,
+      "loss": 0.0126,
+      "step": 4630
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.957074074074075e-05,
+      "loss": 0.0132,
+      "step": 4640
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.956981481481482e-05,
+      "loss": 0.012,
+      "step": 4650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.956888888888889e-05,
+      "loss": 0.0153,
+      "step": 4660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9567962962962965e-05,
+      "loss": 0.009,
+      "step": 4670
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.956703703703704e-05,
+      "loss": 0.0101,
+      "step": 4680
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9566111111111115e-05,
+      "loss": 0.0121,
+      "step": 4690
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.956527777777778e-05,
+      "loss": 0.0112,
+      "step": 4700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9564351851851855e-05,
+      "loss": 0.0137,
+      "step": 4710
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.956342592592593e-05,
+      "loss": 0.0107,
+      "step": 4720
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.95625e-05,
+      "loss": 0.0138,
+      "step": 4730
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9561574074074074e-05,
+      "loss": 0.0154,
+      "step": 4740
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9560648148148155e-05,
+      "loss": 0.013,
+      "step": 4750
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9559722222222224e-05,
+      "loss": 0.0128,
+      "step": 4760
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.95587962962963e-05,
+      "loss": 0.01,
+      "step": 4770
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9557870370370374e-05,
+      "loss": 0.0097,
+      "step": 4780
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.955694444444445e-05,
+      "loss": 0.0134,
+      "step": 4790
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9556018518518524e-05,
+      "loss": 0.0125,
+      "step": 4800
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.955509259259259e-05,
+      "loss": 0.0141,
+      "step": 4810
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.955416666666667e-05,
+      "loss": 0.0083,
+      "step": 4820
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.955324074074074e-05,
+      "loss": 0.0109,
+      "step": 4830
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.955231481481482e-05,
+      "loss": 0.0129,
+      "step": 4840
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9551388888888886e-05,
+      "loss": 0.0079,
+      "step": 4850
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.955046296296297e-05,
+      "loss": 0.0124,
+      "step": 4860
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.954953703703704e-05,
+      "loss": 0.0172,
+      "step": 4870
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.954861111111112e-05,
+      "loss": 0.0115,
+      "step": 4880
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.9547685185185186e-05,
+      "loss": 0.011,
+      "step": 4890
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.954675925925926e-05,
+      "loss": 0.0124,
+      "step": 4900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.9545833333333336e-05,
+      "loss": 0.0118,
+      "step": 4910
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.954490740740741e-05,
+      "loss": 0.0071,
+      "step": 4920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.9543981481481486e-05,
+      "loss": 0.0126,
+      "step": 4930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.9543055555555554e-05,
+      "loss": 0.0129,
+      "step": 4940
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.954212962962963e-05,
+      "loss": 0.0149,
+      "step": 4950
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.9541203703703704e-05,
+      "loss": 0.0108,
+      "step": 4960
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.954027777777778e-05,
+      "loss": 0.0088,
+      "step": 4970
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.9539351851851855e-05,
+      "loss": 0.0106,
+      "step": 4980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.953842592592593e-05,
+      "loss": 0.0136,
+      "step": 4990
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.9537500000000005e-05,
+      "loss": 0.0116,
+      "step": 5000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.953657407407408e-05,
+      "loss": 0.0103,
+      "step": 5010
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.953564814814815e-05,
+      "loss": 0.0146,
+      "step": 5020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.953472222222222e-05,
+      "loss": 0.0113,
+      "step": 5030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.95337962962963e-05,
+      "loss": 0.0137,
+      "step": 5040
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.953287037037037e-05,
+      "loss": 0.0116,
+      "step": 5050
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.953194444444444e-05,
+      "loss": 0.0103,
+      "step": 5060
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.9531018518518517e-05,
+      "loss": 0.0104,
+      "step": 5070
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.95300925925926e-05,
+      "loss": 0.0124,
+      "step": 5080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.9529166666666673e-05,
+      "loss": 0.0132,
+      "step": 5090
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.952824074074074e-05,
+      "loss": 0.0147,
+      "step": 5100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.952731481481482e-05,
+      "loss": 0.0127,
+      "step": 5110
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.952638888888889e-05,
+      "loss": 0.0109,
+      "step": 5120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.952546296296297e-05,
+      "loss": 0.0118,
+      "step": 5130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.952453703703704e-05,
+      "loss": 0.0118,
+      "step": 5140
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.952361111111111e-05,
+      "loss": 0.0101,
+      "step": 5150
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9522685185185185e-05,
+      "loss": 0.0088,
+      "step": 5160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.952175925925926e-05,
+      "loss": 0.0129,
+      "step": 5170
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9520833333333335e-05,
+      "loss": 0.011,
+      "step": 5180
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.951990740740741e-05,
+      "loss": 0.0109,
+      "step": 5190
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9518981481481485e-05,
+      "loss": 0.0098,
+      "step": 5200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.951805555555556e-05,
+      "loss": 0.012,
+      "step": 5210
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9517129629629636e-05,
+      "loss": 0.0082,
+      "step": 5220
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9516203703703704e-05,
+      "loss": 0.0115,
+      "step": 5230
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.951527777777778e-05,
+      "loss": 0.0104,
+      "step": 5240
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9514351851851854e-05,
+      "loss": 0.01,
+      "step": 5250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.951342592592593e-05,
+      "loss": 0.0095,
+      "step": 5260
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.95125e-05,
+      "loss": 0.01,
+      "step": 5270
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.951157407407407e-05,
+      "loss": 0.0096,
+      "step": 5280
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.9510648148148154e-05,
+      "loss": 0.0068,
+      "step": 5290
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.950972222222223e-05,
+      "loss": 0.0108,
+      "step": 5300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.95087962962963e-05,
+      "loss": 0.0092,
+      "step": 5310
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.950787037037037e-05,
+      "loss": 0.008,
+      "step": 5320
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.950694444444445e-05,
+      "loss": 0.0137,
+      "step": 5330
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.950601851851852e-05,
+      "loss": 0.0125,
+      "step": 5340
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.95050925925926e-05,
+      "loss": 0.0107,
+      "step": 5350
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.9504166666666666e-05,
+      "loss": 0.0115,
+      "step": 5360
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.950324074074074e-05,
+      "loss": 0.0109,
+      "step": 5370
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.9502314814814816e-05,
+      "loss": 0.0086,
+      "step": 5380
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.950138888888889e-05,
+      "loss": 0.0113,
+      "step": 5390
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.9500462962962966e-05,
+      "loss": 0.0153,
+      "step": 5400
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.01483889576047659,
+      "eval_runtime": 153.7836,
+      "eval_samples_per_second": 25.946,
+      "eval_steps_per_second": 3.245,
+      "step": 5400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.949953703703704e-05,
+      "loss": 0.0104,
+      "step": 5410
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.9498611111111116e-05,
+      "loss": 0.0115,
+      "step": 5420
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.949768518518519e-05,
+      "loss": 0.0136,
+      "step": 5430
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.949675925925926e-05,
+      "loss": 0.0148,
+      "step": 5440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.9495833333333335e-05,
+      "loss": 0.0106,
+      "step": 5450
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.949490740740741e-05,
+      "loss": 0.012,
+      "step": 5460
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.9493981481481485e-05,
+      "loss": 0.0126,
+      "step": 5470
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.949305555555555e-05,
+      "loss": 0.0101,
+      "step": 5480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.949212962962963e-05,
+      "loss": 0.0123,
+      "step": 5490
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.94912037037037e-05,
+      "loss": 0.0075,
+      "step": 5500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.9490277777777785e-05,
+      "loss": 0.0082,
+      "step": 5510
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.948935185185185e-05,
+      "loss": 0.011,
+      "step": 5520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.948842592592593e-05,
+      "loss": 0.0101,
+      "step": 5530
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.9487500000000003e-05,
+      "loss": 0.0137,
+      "step": 5540
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.948657407407408e-05,
+      "loss": 0.013,
+      "step": 5550
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.9485648148148154e-05,
+      "loss": 0.0131,
+      "step": 5560
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.948472222222222e-05,
+      "loss": 0.0119,
+      "step": 5570
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.94837962962963e-05,
+      "loss": 0.0107,
+      "step": 5580
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.948287037037037e-05,
+      "loss": 0.0102,
+      "step": 5590
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.948194444444445e-05,
+      "loss": 0.0088,
+      "step": 5600
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.9481018518518515e-05,
+      "loss": 0.0088,
+      "step": 5610
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.94800925925926e-05,
+      "loss": 0.0076,
+      "step": 5620
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.947916666666667e-05,
+      "loss": 0.0118,
+      "step": 5630
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.947824074074075e-05,
+      "loss": 0.0068,
+      "step": 5640
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.9477314814814816e-05,
+      "loss": 0.0131,
+      "step": 5650
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.947638888888889e-05,
+      "loss": 0.0141,
+      "step": 5660
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.9475462962962966e-05,
+      "loss": 0.0114,
+      "step": 5670
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.947453703703704e-05,
+      "loss": 0.0124,
+      "step": 5680
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.947361111111111e-05,
+      "loss": 0.0094,
+      "step": 5690
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.9472685185185184e-05,
+      "loss": 0.0167,
+      "step": 5700
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.947175925925926e-05,
+      "loss": 0.0109,
+      "step": 5710
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.947083333333334e-05,
+      "loss": 0.0137,
+      "step": 5720
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.946990740740741e-05,
+      "loss": 0.0098,
+      "step": 5730
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.9468981481481484e-05,
+      "loss": 0.0112,
+      "step": 5740
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.946805555555556e-05,
+      "loss": 0.0117,
+      "step": 5750
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.9467129629629634e-05,
+      "loss": 0.0117,
+      "step": 5760
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.946620370370371e-05,
+      "loss": 0.0121,
+      "step": 5770
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.946527777777778e-05,
+      "loss": 0.0115,
+      "step": 5780
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.946435185185185e-05,
+      "loss": 0.0093,
+      "step": 5790
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.946342592592593e-05,
+      "loss": 0.0085,
+      "step": 5800
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.94625e-05,
+      "loss": 0.006,
+      "step": 5810
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.946157407407407e-05,
+      "loss": 0.0086,
+      "step": 5820
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.946064814814815e-05,
+      "loss": 0.0126,
+      "step": 5830
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.945972222222223e-05,
+      "loss": 0.0115,
+      "step": 5840
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.94587962962963e-05,
+      "loss": 0.0113,
+      "step": 5850
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.945787037037037e-05,
+      "loss": 0.0091,
+      "step": 5860
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.9456944444444446e-05,
+      "loss": 0.0108,
+      "step": 5870
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.945601851851852e-05,
+      "loss": 0.0115,
+      "step": 5880
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.9455092592592597e-05,
+      "loss": 0.0071,
+      "step": 5890
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.9454166666666665e-05,
+      "loss": 0.0085,
+      "step": 5900
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.945324074074074e-05,
+      "loss": 0.0102,
+      "step": 5910
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.9452314814814815e-05,
+      "loss": 0.008,
+      "step": 5920
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.945138888888889e-05,
+      "loss": 0.011,
+      "step": 5930
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.9450462962962965e-05,
+      "loss": 0.0126,
+      "step": 5940
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.944953703703704e-05,
+      "loss": 0.0132,
+      "step": 5950
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.9448611111111115e-05,
+      "loss": 0.0125,
+      "step": 5960
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.944768518518519e-05,
+      "loss": 0.0091,
+      "step": 5970
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.9446759259259265e-05,
+      "loss": 0.0098,
+      "step": 5980
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.9445833333333334e-05,
+      "loss": 0.009,
+      "step": 5990
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.944490740740741e-05,
+      "loss": 0.0097,
+      "step": 6000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.9443981481481484e-05,
+      "loss": 0.0101,
+      "step": 6010
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.944305555555556e-05,
+      "loss": 0.0106,
+      "step": 6020
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.944212962962963e-05,
+      "loss": 0.0077,
+      "step": 6030
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.94412037037037e-05,
+      "loss": 0.0143,
+      "step": 6040
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.9440277777777784e-05,
+      "loss": 0.01,
+      "step": 6050
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.943935185185186e-05,
+      "loss": 0.0081,
+      "step": 6060
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.943842592592593e-05,
+      "loss": 0.0127,
+      "step": 6070
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.94375e-05,
+      "loss": 0.0109,
+      "step": 6080
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.943657407407408e-05,
+      "loss": 0.0088,
+      "step": 6090
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.943564814814815e-05,
+      "loss": 0.0104,
+      "step": 6100
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.943472222222222e-05,
+      "loss": 0.0096,
+      "step": 6110
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.9433796296296296e-05,
+      "loss": 0.0081,
+      "step": 6120
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.943287037037037e-05,
+      "loss": 0.0123,
+      "step": 6130
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.9431944444444446e-05,
+      "loss": 0.0082,
+      "step": 6140
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.943101851851852e-05,
+      "loss": 0.0115,
+      "step": 6150
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.9430092592592596e-05,
+      "loss": 0.009,
+      "step": 6160
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.942916666666667e-05,
+      "loss": 0.0099,
+      "step": 6170
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.9428240740740746e-05,
+      "loss": 0.0114,
+      "step": 6180
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.942731481481482e-05,
+      "loss": 0.0094,
+      "step": 6190
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.942638888888889e-05,
+      "loss": 0.0115,
+      "step": 6200
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.9425462962962964e-05,
+      "loss": 0.0128,
+      "step": 6210
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.942453703703704e-05,
+      "loss": 0.0126,
+      "step": 6220
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.9423611111111115e-05,
+      "loss": 0.0117,
+      "step": 6230
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.942268518518518e-05,
+      "loss": 0.0092,
+      "step": 6240
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.942175925925926e-05,
+      "loss": 0.0113,
+      "step": 6250
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.942083333333334e-05,
+      "loss": 0.0087,
+      "step": 6260
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.9419907407407415e-05,
+      "loss": 0.0111,
+      "step": 6270
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.941898148148148e-05,
+      "loss": 0.0098,
+      "step": 6280
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.941805555555556e-05,
+      "loss": 0.0118,
+      "step": 6290
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.941712962962963e-05,
+      "loss": 0.0101,
+      "step": 6300
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.941620370370371e-05,
+      "loss": 0.0114,
+      "step": 6310
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.9415277777777777e-05,
+      "loss": 0.0075,
+      "step": 6320
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.941435185185185e-05,
+      "loss": 0.0124,
+      "step": 6330
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.941342592592593e-05,
+      "loss": 0.0101,
+      "step": 6340
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.94125e-05,
+      "loss": 0.0084,
+      "step": 6350
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.941157407407408e-05,
+      "loss": 0.0136,
+      "step": 6360
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.941064814814815e-05,
+      "loss": 0.0077,
+      "step": 6370
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.940972222222223e-05,
+      "loss": 0.0088,
+      "step": 6380
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.94087962962963e-05,
+      "loss": 0.0113,
+      "step": 6390
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.940787037037038e-05,
+      "loss": 0.0084,
+      "step": 6400
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.9406944444444445e-05,
+      "loss": 0.008,
+      "step": 6410
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.940601851851852e-05,
+      "loss": 0.0119,
+      "step": 6420
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.9405092592592595e-05,
+      "loss": 0.0107,
+      "step": 6430
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.940416666666667e-05,
+      "loss": 0.0159,
+      "step": 6440
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.940324074074074e-05,
+      "loss": 0.0127,
+      "step": 6450
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.9402314814814814e-05,
+      "loss": 0.012,
+      "step": 6460
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.940138888888889e-05,
+      "loss": 0.0098,
+      "step": 6470
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.940046296296297e-05,
+      "loss": 0.0097,
+      "step": 6480
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.939953703703704e-05,
+      "loss": 0.0089,
+      "step": 6490
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.9398611111111114e-05,
+      "loss": 0.0076,
+      "step": 6500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.939768518518519e-05,
+      "loss": 0.0105,
+      "step": 6510
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.9396759259259264e-05,
+      "loss": 0.0093,
+      "step": 6520
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.939583333333333e-05,
+      "loss": 0.0105,
+      "step": 6530
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.939490740740741e-05,
+      "loss": 0.0093,
+      "step": 6540
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.939398148148148e-05,
+      "loss": 0.0075,
+      "step": 6550
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.939305555555556e-05,
+      "loss": 0.007,
+      "step": 6560
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.939212962962963e-05,
+      "loss": 0.0117,
+      "step": 6570
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.93912037037037e-05,
+      "loss": 0.0095,
+      "step": 6580
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.939027777777778e-05,
+      "loss": 0.0095,
+      "step": 6590
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.938935185185186e-05,
+      "loss": 0.0079,
+      "step": 6600
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.938842592592593e-05,
+      "loss": 0.0122,
+      "step": 6610
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.93875e-05,
+      "loss": 0.0107,
+      "step": 6620
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.9386574074074076e-05,
+      "loss": 0.0105,
+      "step": 6630
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.938564814814815e-05,
+      "loss": 0.0104,
+      "step": 6640
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.9384722222222226e-05,
+      "loss": 0.0104,
+      "step": 6650
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.9383796296296295e-05,
+      "loss": 0.0112,
+      "step": 6660
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.938287037037037e-05,
+      "loss": 0.0127,
+      "step": 6670
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.9381944444444445e-05,
+      "loss": 0.0104,
+      "step": 6680
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.938101851851852e-05,
+      "loss": 0.0106,
+      "step": 6690
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.9380092592592595e-05,
+      "loss": 0.0082,
+      "step": 6700
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.937916666666667e-05,
+      "loss": 0.008,
+      "step": 6710
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.9378240740740745e-05,
+      "loss": 0.012,
+      "step": 6720
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.937731481481482e-05,
+      "loss": 0.0112,
+      "step": 6730
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.937638888888889e-05,
+      "loss": 0.0106,
+      "step": 6740
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.937546296296296e-05,
+      "loss": 0.0095,
+      "step": 6750
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.937453703703704e-05,
+      "loss": 0.0099,
+      "step": 6760
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.937361111111111e-05,
+      "loss": 0.014,
+      "step": 6770
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.937268518518519e-05,
+      "loss": 0.0132,
+      "step": 6780
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.937175925925926e-05,
+      "loss": 0.01,
+      "step": 6790
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.937083333333334e-05,
+      "loss": 0.0092,
+      "step": 6800
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.9369907407407414e-05,
+      "loss": 0.0131,
+      "step": 6810
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.936898148148149e-05,
+      "loss": 0.0087,
+      "step": 6820
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.936805555555556e-05,
+      "loss": 0.0126,
+      "step": 6830
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.936712962962963e-05,
+      "loss": 0.0084,
+      "step": 6840
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.936620370370371e-05,
+      "loss": 0.0109,
+      "step": 6850
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.936527777777778e-05,
+      "loss": 0.0123,
+      "step": 6860
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.936435185185185e-05,
+      "loss": 0.0098,
+      "step": 6870
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.9363425925925925e-05,
+      "loss": 0.0102,
+      "step": 6880
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.93625e-05,
+      "loss": 0.0087,
+      "step": 6890
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.9361574074074076e-05,
+      "loss": 0.0086,
+      "step": 6900
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.936064814814815e-05,
+      "loss": 0.0086,
+      "step": 6910
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.9359722222222226e-05,
+      "loss": 0.0085,
+      "step": 6920
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.93587962962963e-05,
+      "loss": 0.0085,
+      "step": 6930
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.9357870370370376e-05,
+      "loss": 0.0069,
+      "step": 6940
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.9356944444444444e-05,
+      "loss": 0.0098,
+      "step": 6950
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.935601851851852e-05,
+      "loss": 0.009,
+      "step": 6960
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.9355092592592594e-05,
+      "loss": 0.0081,
+      "step": 6970
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.935416666666667e-05,
+      "loss": 0.0152,
+      "step": 6980
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.9353240740740744e-05,
+      "loss": 0.012,
+      "step": 6990
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.935231481481481e-05,
+      "loss": 0.0132,
+      "step": 7000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.935138888888889e-05,
+      "loss": 0.0127,
+      "step": 7010
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.935046296296297e-05,
+      "loss": 0.0087,
+      "step": 7020
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.9349537037037044e-05,
+      "loss": 0.0084,
+      "step": 7030
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.934861111111111e-05,
+      "loss": 0.0091,
+      "step": 7040
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.934768518518519e-05,
+      "loss": 0.0077,
+      "step": 7050
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.934675925925926e-05,
+      "loss": 0.0101,
+      "step": 7060
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.934583333333334e-05,
+      "loss": 0.0091,
+      "step": 7070
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.9344907407407406e-05,
+      "loss": 0.0093,
+      "step": 7080
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.934398148148148e-05,
+      "loss": 0.007,
+      "step": 7090
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.9343055555555556e-05,
+      "loss": 0.0066,
+      "step": 7100
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.934212962962963e-05,
+      "loss": 0.0106,
+      "step": 7110
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.9341203703703706e-05,
+      "loss": 0.0133,
+      "step": 7120
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.934027777777778e-05,
+      "loss": 0.0087,
+      "step": 7130
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.9339351851851857e-05,
+      "loss": 0.0109,
+      "step": 7140
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.933842592592593e-05,
+      "loss": 0.0109,
+      "step": 7150
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.93375e-05,
+      "loss": 0.0086,
+      "step": 7160
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.9336574074074075e-05,
+      "loss": 0.0115,
+      "step": 7170
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.933564814814815e-05,
+      "loss": 0.0073,
+      "step": 7180
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.9334722222222225e-05,
+      "loss": 0.0075,
+      "step": 7190
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.93337962962963e-05,
+      "loss": 0.011,
+      "step": 7200
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.933287037037037e-05,
+      "loss": 0.0095,
+      "step": 7210
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.9331944444444443e-05,
+      "loss": 0.0102,
+      "step": 7220
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.9331018518518525e-05,
+      "loss": 0.0087,
+      "step": 7230
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.93300925925926e-05,
+      "loss": 0.0068,
+      "step": 7240
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.932916666666667e-05,
+      "loss": 0.0104,
+      "step": 7250
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.9328240740740744e-05,
+      "loss": 0.0095,
+      "step": 7260
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.932731481481482e-05,
+      "loss": 0.0085,
+      "step": 7270
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.9326388888888894e-05,
+      "loss": 0.0096,
+      "step": 7280
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.932546296296296e-05,
+      "loss": 0.0095,
+      "step": 7290
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.932453703703704e-05,
+      "loss": 0.0105,
+      "step": 7300
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.932361111111111e-05,
+      "loss": 0.0125,
+      "step": 7310
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.932268518518519e-05,
+      "loss": 0.0067,
+      "step": 7320
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.932175925925926e-05,
+      "loss": 0.0068,
+      "step": 7330
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.932083333333334e-05,
+      "loss": 0.0103,
+      "step": 7340
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.931990740740741e-05,
+      "loss": 0.0104,
+      "step": 7350
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.931898148148149e-05,
+      "loss": 0.0079,
+      "step": 7360
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9318055555555556e-05,
+      "loss": 0.009,
+      "step": 7370
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.931712962962963e-05,
+      "loss": 0.0082,
+      "step": 7380
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.9316203703703706e-05,
+      "loss": 0.0114,
+      "step": 7390
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.931527777777778e-05,
+      "loss": 0.01,
+      "step": 7400
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.9314351851851856e-05,
+      "loss": 0.0077,
+      "step": 7410
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.9313425925925924e-05,
+      "loss": 0.0103,
+      "step": 7420
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.93125e-05,
+      "loss": 0.0126,
+      "step": 7430
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.9311574074074074e-05,
+      "loss": 0.0069,
+      "step": 7440
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.9310648148148156e-05,
+      "loss": 0.0108,
+      "step": 7450
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.9309722222222224e-05,
+      "loss": 0.009,
+      "step": 7460
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.93087962962963e-05,
+      "loss": 0.0094,
+      "step": 7470
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.9307870370370375e-05,
+      "loss": 0.0061,
+      "step": 7480
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.930694444444445e-05,
+      "loss": 0.0101,
+      "step": 7490
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.930601851851852e-05,
+      "loss": 0.007,
+      "step": 7500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.930509259259259e-05,
+      "loss": 0.0105,
+      "step": 7510
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.930416666666667e-05,
+      "loss": 0.0087,
+      "step": 7520
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.930324074074074e-05,
+      "loss": 0.0113,
+      "step": 7530
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.930231481481482e-05,
+      "loss": 0.0099,
+      "step": 7540
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.9301388888888886e-05,
+      "loss": 0.0089,
+      "step": 7550
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.930046296296297e-05,
+      "loss": 0.0133,
+      "step": 7560
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.929953703703704e-05,
+      "loss": 0.0133,
+      "step": 7570
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.929861111111111e-05,
+      "loss": 0.0084,
+      "step": 7580
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.9297685185185187e-05,
+      "loss": 0.0106,
+      "step": 7590
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.929675925925926e-05,
+      "loss": 0.0082,
+      "step": 7600
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.929583333333334e-05,
+      "loss": 0.0099,
+      "step": 7610
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.929490740740741e-05,
+      "loss": 0.006,
+      "step": 7620
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.929398148148148e-05,
+      "loss": 0.0086,
+      "step": 7630
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.9293055555555555e-05,
+      "loss": 0.0087,
+      "step": 7640
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.929212962962963e-05,
+      "loss": 0.0102,
+      "step": 7650
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.9291203703703705e-05,
+      "loss": 0.0134,
+      "step": 7660
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.929027777777778e-05,
+      "loss": 0.0073,
+      "step": 7670
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.9289351851851855e-05,
+      "loss": 0.0108,
+      "step": 7680
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.928842592592593e-05,
+      "loss": 0.0134,
+      "step": 7690
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.9287500000000005e-05,
+      "loss": 0.0115,
+      "step": 7700
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.9286574074074074e-05,
+      "loss": 0.0099,
+      "step": 7710
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.928564814814815e-05,
+      "loss": 0.0114,
+      "step": 7720
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.9284722222222224e-05,
+      "loss": 0.0115,
+      "step": 7730
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.92837962962963e-05,
+      "loss": 0.0107,
+      "step": 7740
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.9282870370370374e-05,
+      "loss": 0.0061,
+      "step": 7750
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.928194444444444e-05,
+      "loss": 0.0066,
+      "step": 7760
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.9281018518518524e-05,
+      "loss": 0.006,
+      "step": 7770
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.928018518518518e-05,
+      "loss": 0.0108,
+      "step": 7780
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.9279259259259264e-05,
+      "loss": 0.0119,
+      "step": 7790
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.927833333333334e-05,
+      "loss": 0.0088,
+      "step": 7800
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.9277407407407414e-05,
+      "loss": 0.0052,
+      "step": 7810
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.927648148148148e-05,
+      "loss": 0.0108,
+      "step": 7820
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.927555555555556e-05,
+      "loss": 0.0096,
+      "step": 7830
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.927462962962963e-05,
+      "loss": 0.0107,
+      "step": 7840
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.927370370370371e-05,
+      "loss": 0.0098,
+      "step": 7850
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.9272777777777776e-05,
+      "loss": 0.008,
+      "step": 7860
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.927185185185185e-05,
+      "loss": 0.011,
+      "step": 7870
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.9270925925925926e-05,
+      "loss": 0.0097,
+      "step": 7880
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.927000000000001e-05,
+      "loss": 0.0101,
+      "step": 7890
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.9269074074074076e-05,
+      "loss": 0.0122,
+      "step": 7900
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.926814814814815e-05,
+      "loss": 0.0068,
+      "step": 7910
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.9267222222222226e-05,
+      "loss": 0.0105,
+      "step": 7920
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.92662962962963e-05,
+      "loss": 0.0102,
+      "step": 7930
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.9265370370370376e-05,
+      "loss": 0.0065,
+      "step": 7940
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.9264444444444444e-05,
+      "loss": 0.0065,
+      "step": 7950
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.926351851851852e-05,
+      "loss": 0.0098,
+      "step": 7960
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.9262592592592595e-05,
+      "loss": 0.0081,
+      "step": 7970
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.926166666666667e-05,
+      "loss": 0.0129,
+      "step": 7980
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.926074074074074e-05,
+      "loss": 0.0097,
+      "step": 7990
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.925981481481482e-05,
+      "loss": 0.0085,
+      "step": 8000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.9258888888888895e-05,
+      "loss": 0.0069,
+      "step": 8010
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.925796296296297e-05,
+      "loss": 0.0109,
+      "step": 8020
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.925703703703704e-05,
+      "loss": 0.0088,
+      "step": 8030
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.925611111111111e-05,
+      "loss": 0.0074,
+      "step": 8040
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.925518518518519e-05,
+      "loss": 0.0108,
+      "step": 8050
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.925425925925926e-05,
+      "loss": 0.0101,
+      "step": 8060
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.925333333333333e-05,
+      "loss": 0.01,
+      "step": 8070
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.925240740740741e-05,
+      "loss": 0.0107,
+      "step": 8080
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.925148148148148e-05,
+      "loss": 0.0115,
+      "step": 8090
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.925055555555556e-05,
+      "loss": 0.0091,
+      "step": 8100
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.924962962962963e-05,
+      "loss": 0.0093,
+      "step": 8110
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.924870370370371e-05,
+      "loss": 0.0109,
+      "step": 8120
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.924777777777778e-05,
+      "loss": 0.0073,
+      "step": 8130
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.924685185185186e-05,
+      "loss": 0.0112,
+      "step": 8140
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.924592592592593e-05,
+      "loss": 0.0073,
+      "step": 8150
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.9245e-05,
+      "loss": 0.0092,
+      "step": 8160
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.9244074074074075e-05,
+      "loss": 0.0077,
+      "step": 8170
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.924314814814815e-05,
+      "loss": 0.0088,
+      "step": 8180
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.9242222222222225e-05,
+      "loss": 0.01,
+      "step": 8190
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.9241296296296294e-05,
+      "loss": 0.0101,
+      "step": 8200
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.924037037037037e-05,
+      "loss": 0.0109,
+      "step": 8210
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.923944444444445e-05,
+      "loss": 0.0127,
+      "step": 8220
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.9238518518518526e-05,
+      "loss": 0.008,
+      "step": 8230
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.9237592592592594e-05,
+      "loss": 0.0098,
+      "step": 8240
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.923666666666667e-05,
+      "loss": 0.0073,
+      "step": 8250
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.9235740740740744e-05,
+      "loss": 0.0084,
+      "step": 8260
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.923481481481482e-05,
+      "loss": 0.01,
+      "step": 8270
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.923388888888889e-05,
+      "loss": 0.0102,
+      "step": 8280
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.923296296296296e-05,
+      "loss": 0.0104,
+      "step": 8290
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.923203703703704e-05,
+      "loss": 0.0107,
+      "step": 8300
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.923111111111111e-05,
+      "loss": 0.0062,
+      "step": 8310
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.923018518518519e-05,
+      "loss": 0.0124,
+      "step": 8320
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.922925925925926e-05,
+      "loss": 0.0085,
+      "step": 8330
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.922833333333334e-05,
+      "loss": 0.009,
+      "step": 8340
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.922740740740741e-05,
+      "loss": 0.0107,
+      "step": 8350
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.922648148148149e-05,
+      "loss": 0.0089,
+      "step": 8360
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.9225555555555556e-05,
+      "loss": 0.008,
+      "step": 8370
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.922462962962963e-05,
+      "loss": 0.0109,
+      "step": 8380
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.9223703703703706e-05,
+      "loss": 0.0107,
+      "step": 8390
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.922277777777778e-05,
+      "loss": 0.0116,
+      "step": 8400
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.922185185185185e-05,
+      "loss": 0.0068,
+      "step": 8410
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.9220925925925925e-05,
+      "loss": 0.0075,
+      "step": 8420
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.9220000000000006e-05,
+      "loss": 0.0064,
+      "step": 8430
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.921907407407408e-05,
+      "loss": 0.0106,
+      "step": 8440
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.921814814814815e-05,
+      "loss": 0.0071,
+      "step": 8450
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.9217222222222225e-05,
+      "loss": 0.0074,
+      "step": 8460
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.92162962962963e-05,
+      "loss": 0.0097,
+      "step": 8470
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.9215370370370375e-05,
+      "loss": 0.0088,
+      "step": 8480
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.921444444444444e-05,
+      "loss": 0.0093,
+      "step": 8490
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.921351851851852e-05,
+      "loss": 0.008,
+      "step": 8500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.921259259259259e-05,
+      "loss": 0.0086,
+      "step": 8510
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.921166666666667e-05,
+      "loss": 0.0099,
+      "step": 8520
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.9210740740740743e-05,
+      "loss": 0.009,
+      "step": 8530
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.920981481481482e-05,
+      "loss": 0.0085,
+      "step": 8540
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.9208888888888894e-05,
+      "loss": 0.0108,
+      "step": 8550
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.920796296296297e-05,
+      "loss": 0.012,
+      "step": 8560
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.9207037037037044e-05,
+      "loss": 0.0061,
+      "step": 8570
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.920611111111111e-05,
+      "loss": 0.007,
+      "step": 8580
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.920518518518519e-05,
+      "loss": 0.0082,
+      "step": 8590
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.920425925925926e-05,
+      "loss": 0.0102,
+      "step": 8600
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.920333333333334e-05,
+      "loss": 0.0111,
+      "step": 8610
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.9202407407407405e-05,
+      "loss": 0.0123,
+      "step": 8620
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.920148148148148e-05,
+      "loss": 0.0102,
+      "step": 8630
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.9200555555555556e-05,
+      "loss": 0.01,
+      "step": 8640
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.919962962962964e-05,
+      "loss": 0.0094,
+      "step": 8650
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.9198703703703706e-05,
+      "loss": 0.0097,
+      "step": 8660
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.919777777777778e-05,
+      "loss": 0.0067,
+      "step": 8670
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.9196851851851856e-05,
+      "loss": 0.011,
+      "step": 8680
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.919592592592593e-05,
+      "loss": 0.0122,
+      "step": 8690
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.9195e-05,
+      "loss": 0.0133,
+      "step": 8700
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.9194074074074074e-05,
+      "loss": 0.0093,
+      "step": 8710
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.919314814814815e-05,
+      "loss": 0.0086,
+      "step": 8720
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.9192222222222224e-05,
+      "loss": 0.0089,
+      "step": 8730
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.91912962962963e-05,
+      "loss": 0.0095,
+      "step": 8740
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.919037037037037e-05,
+      "loss": 0.0092,
+      "step": 8750
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.918944444444445e-05,
+      "loss": 0.0073,
+      "step": 8760
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.9188518518518524e-05,
+      "loss": 0.0054,
+      "step": 8770
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.91875925925926e-05,
+      "loss": 0.0085,
+      "step": 8780
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.918666666666667e-05,
+      "loss": 0.009,
+      "step": 8790
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.918574074074074e-05,
+      "loss": 0.005,
+      "step": 8800
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.918481481481482e-05,
+      "loss": 0.01,
+      "step": 8810
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.918388888888889e-05,
+      "loss": 0.0068,
+      "step": 8820
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.918296296296296e-05,
+      "loss": 0.0078,
+      "step": 8830
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.9182037037037036e-05,
+      "loss": 0.0066,
+      "step": 8840
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.918111111111111e-05,
+      "loss": 0.0113,
+      "step": 8850
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.918018518518519e-05,
+      "loss": 0.0106,
+      "step": 8860
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.917925925925926e-05,
+      "loss": 0.0086,
+      "step": 8870
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.9178333333333337e-05,
+      "loss": 0.0052,
+      "step": 8880
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.917740740740741e-05,
+      "loss": 0.0079,
+      "step": 8890
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.917648148148149e-05,
+      "loss": 0.0079,
+      "step": 8900
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.9175555555555555e-05,
+      "loss": 0.0083,
+      "step": 8910
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.917462962962963e-05,
+      "loss": 0.009,
+      "step": 8920
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.9173703703703705e-05,
+      "loss": 0.0104,
+      "step": 8930
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.917277777777778e-05,
+      "loss": 0.0095,
+      "step": 8940
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.9171851851851855e-05,
+      "loss": 0.0114,
+      "step": 8950
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.9170925925925923e-05,
+      "loss": 0.013,
+      "step": 8960
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.9170000000000005e-05,
+      "loss": 0.0094,
+      "step": 8970
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.916907407407408e-05,
+      "loss": 0.0101,
+      "step": 8980
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.9168148148148155e-05,
+      "loss": 0.009,
+      "step": 8990
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.9167222222222224e-05,
+      "loss": 0.0084,
+      "step": 9000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.91662962962963e-05,
+      "loss": 0.0079,
+      "step": 9010
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.9165370370370374e-05,
+      "loss": 0.0099,
+      "step": 9020
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.916444444444445e-05,
+      "loss": 0.008,
+      "step": 9030
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.916351851851852e-05,
+      "loss": 0.0086,
+      "step": 9040
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.916259259259259e-05,
+      "loss": 0.0118,
+      "step": 9050
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.916166666666667e-05,
+      "loss": 0.0102,
+      "step": 9060
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.916074074074074e-05,
+      "loss": 0.0059,
+      "step": 9070
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.915981481481482e-05,
+      "loss": 0.0109,
+      "step": 9080
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.915888888888889e-05,
+      "loss": 0.0119,
+      "step": 9090
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.915796296296297e-05,
+      "loss": 0.0091,
+      "step": 9100
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.915703703703704e-05,
+      "loss": 0.01,
+      "step": 9110
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.915611111111111e-05,
+      "loss": 0.0085,
+      "step": 9120
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.9155185185185186e-05,
+      "loss": 0.009,
+      "step": 9130
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.915425925925926e-05,
+      "loss": 0.0095,
+      "step": 9140
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.9153333333333336e-05,
+      "loss": 0.0102,
+      "step": 9150
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.915240740740741e-05,
+      "loss": 0.0077,
+      "step": 9160
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.915148148148148e-05,
+      "loss": 0.0088,
+      "step": 9170
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.9150555555555554e-05,
+      "loss": 0.0126,
+      "step": 9180
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.9149629629629636e-05,
+      "loss": 0.009,
+      "step": 9190
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.914870370370371e-05,
+      "loss": 0.0096,
+      "step": 9200
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.914777777777778e-05,
+      "loss": 0.0065,
+      "step": 9210
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.9146851851851855e-05,
+      "loss": 0.0125,
+      "step": 9220
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.914592592592593e-05,
+      "loss": 0.008,
+      "step": 9230
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.9145000000000005e-05,
+      "loss": 0.0079,
+      "step": 9240
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.914407407407407e-05,
+      "loss": 0.0063,
+      "step": 9250
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.914314814814815e-05,
+      "loss": 0.0112,
+      "step": 9260
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.914222222222222e-05,
+      "loss": 0.0084,
+      "step": 9270
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.91412962962963e-05,
+      "loss": 0.0083,
+      "step": 9280
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.914037037037037e-05,
+      "loss": 0.0113,
+      "step": 9290
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.913944444444445e-05,
+      "loss": 0.0091,
+      "step": 9300
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.913851851851852e-05,
+      "loss": 0.0092,
+      "step": 9310
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.91375925925926e-05,
+      "loss": 0.0118,
+      "step": 9320
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.9136666666666667e-05,
+      "loss": 0.0104,
+      "step": 9330
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.913574074074074e-05,
+      "loss": 0.0098,
+      "step": 9340
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.913481481481482e-05,
+      "loss": 0.0073,
+      "step": 9350
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.913388888888889e-05,
+      "loss": 0.0061,
+      "step": 9360
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.913296296296297e-05,
+      "loss": 0.0072,
+      "step": 9370
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.9132037037037035e-05,
+      "loss": 0.0124,
+      "step": 9380
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.913111111111111e-05,
+      "loss": 0.0102,
+      "step": 9390
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.913018518518519e-05,
+      "loss": 0.0062,
+      "step": 9400
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.912925925925927e-05,
+      "loss": 0.012,
+      "step": 9410
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.9128333333333335e-05,
+      "loss": 0.0078,
+      "step": 9420
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.912740740740741e-05,
+      "loss": 0.0061,
+      "step": 9430
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.9126481481481485e-05,
+      "loss": 0.0131,
+      "step": 9440
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.912555555555556e-05,
+      "loss": 0.0091,
+      "step": 9450
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.912462962962963e-05,
+      "loss": 0.0068,
+      "step": 9460
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.9123703703703704e-05,
+      "loss": 0.0048,
+      "step": 9470
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.912277777777778e-05,
+      "loss": 0.007,
+      "step": 9480
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.9121851851851854e-05,
+      "loss": 0.0071,
+      "step": 9490
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.912092592592593e-05,
+      "loss": 0.0105,
+      "step": 9500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.9120000000000004e-05,
+      "loss": 0.0121,
+      "step": 9510
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.911907407407408e-05,
+      "loss": 0.0069,
+      "step": 9520
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.9118148148148154e-05,
+      "loss": 0.008,
+      "step": 9530
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.911722222222222e-05,
+      "loss": 0.0096,
+      "step": 9540
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.91162962962963e-05,
+      "loss": 0.0078,
+      "step": 9550
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.911537037037037e-05,
+      "loss": 0.0076,
+      "step": 9560
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.911444444444445e-05,
+      "loss": 0.0109,
+      "step": 9570
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.911351851851852e-05,
+      "loss": 0.007,
+      "step": 9580
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.911259259259259e-05,
+      "loss": 0.0102,
+      "step": 9590
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.9111666666666666e-05,
+      "loss": 0.0079,
+      "step": 9600
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.911074074074074e-05,
+      "loss": 0.0069,
+      "step": 9610
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.910981481481482e-05,
+      "loss": 0.0122,
+      "step": 9620
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.910888888888889e-05,
+      "loss": 0.0096,
+      "step": 9630
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.9107962962962966e-05,
+      "loss": 0.0102,
+      "step": 9640
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.910703703703704e-05,
+      "loss": 0.0121,
+      "step": 9650
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.9106111111111116e-05,
+      "loss": 0.0099,
+      "step": 9660
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.9105185185185185e-05,
+      "loss": 0.0098,
+      "step": 9670
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.910425925925926e-05,
+      "loss": 0.0084,
+      "step": 9680
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.9103333333333335e-05,
+      "loss": 0.0092,
+      "step": 9690
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.910240740740741e-05,
+      "loss": 0.0104,
+      "step": 9700
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.9101481481481485e-05,
+      "loss": 0.0086,
+      "step": 9710
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.910055555555555e-05,
+      "loss": 0.0076,
+      "step": 9720
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.9099629629629635e-05,
+      "loss": 0.0072,
+      "step": 9730
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.909870370370371e-05,
+      "loss": 0.0092,
+      "step": 9740
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.909777777777778e-05,
+      "loss": 0.0059,
+      "step": 9750
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.909685185185185e-05,
+      "loss": 0.0115,
+      "step": 9760
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.909592592592593e-05,
+      "loss": 0.0073,
+      "step": 9770
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.9095000000000003e-05,
+      "loss": 0.0074,
+      "step": 9780
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.909407407407408e-05,
+      "loss": 0.0107,
+      "step": 9790
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.909314814814815e-05,
+      "loss": 0.0104,
+      "step": 9800
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.909222222222222e-05,
+      "loss": 0.0082,
+      "step": 9810
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.90912962962963e-05,
+      "loss": 0.0123,
+      "step": 9820
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.909037037037038e-05,
+      "loss": 0.0099,
+      "step": 9830
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.908944444444445e-05,
+      "loss": 0.0085,
+      "step": 9840
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.908851851851852e-05,
+      "loss": 0.0089,
+      "step": 9850
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.90875925925926e-05,
+      "loss": 0.0079,
+      "step": 9860
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.908666666666667e-05,
+      "loss": 0.0111,
+      "step": 9870
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.908574074074074e-05,
+      "loss": 0.0084,
+      "step": 9880
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.9084814814814815e-05,
+      "loss": 0.0096,
+      "step": 9890
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.908388888888889e-05,
+      "loss": 0.0078,
+      "step": 9900
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.9082962962962966e-05,
+      "loss": 0.0069,
+      "step": 9910
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.908203703703704e-05,
+      "loss": 0.008,
+      "step": 9920
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.908111111111111e-05,
+      "loss": 0.0069,
+      "step": 9930
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.908018518518519e-05,
+      "loss": 0.0056,
+      "step": 9940
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.9079259259259266e-05,
+      "loss": 0.0061,
+      "step": 9950
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.9078333333333334e-05,
+      "loss": 0.0092,
+      "step": 9960
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.907740740740741e-05,
+      "loss": 0.0064,
+      "step": 9970
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.9076481481481484e-05,
+      "loss": 0.0069,
+      "step": 9980
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.907555555555556e-05,
+      "loss": 0.0139,
+      "step": 9990
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.9074629629629634e-05,
+      "loss": 0.0091,
+      "step": 10000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.90737037037037e-05,
+      "loss": 0.0094,
+      "step": 10010
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.907277777777778e-05,
+      "loss": 0.0096,
+      "step": 10020
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.907185185185185e-05,
+      "loss": 0.0094,
+      "step": 10030
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.907092592592593e-05,
+      "loss": 0.0098,
+      "step": 10040
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.907e-05,
+      "loss": 0.0086,
+      "step": 10050
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.906907407407408e-05,
+      "loss": 0.0117,
+      "step": 10060
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.906814814814815e-05,
+      "loss": 0.0071,
+      "step": 10070
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.906722222222223e-05,
+      "loss": 0.007,
+      "step": 10080
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.9066296296296296e-05,
+      "loss": 0.0075,
+      "step": 10090
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.906537037037037e-05,
+      "loss": 0.0086,
+      "step": 10100
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.9064444444444446e-05,
+      "loss": 0.009,
+      "step": 10110
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.906351851851852e-05,
+      "loss": 0.0076,
+      "step": 10120
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.9062592592592596e-05,
+      "loss": 0.0101,
+      "step": 10130
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.9061666666666665e-05,
+      "loss": 0.0133,
+      "step": 10140
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.906074074074074e-05,
+      "loss": 0.0058,
+      "step": 10150
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.905981481481482e-05,
+      "loss": 0.0079,
+      "step": 10160
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.905888888888889e-05,
+      "loss": 0.0065,
+      "step": 10170
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.9057962962962965e-05,
+      "loss": 0.0073,
+      "step": 10180
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.905703703703704e-05,
+      "loss": 0.0091,
+      "step": 10190
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.9056111111111115e-05,
+      "loss": 0.011,
+      "step": 10200
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.905518518518519e-05,
+      "loss": 0.0101,
+      "step": 10210
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.905425925925926e-05,
+      "loss": 0.0114,
+      "step": 10220
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.9053333333333333e-05,
+      "loss": 0.0111,
+      "step": 10230
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.905240740740741e-05,
+      "loss": 0.0107,
+      "step": 10240
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.9051481481481484e-05,
+      "loss": 0.0115,
+      "step": 10250
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.905055555555555e-05,
+      "loss": 0.0092,
+      "step": 10260
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.9049629629629634e-05,
+      "loss": 0.0088,
+      "step": 10270
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.904870370370371e-05,
+      "loss": 0.0071,
+      "step": 10280
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.9047777777777784e-05,
+      "loss": 0.0065,
+      "step": 10290
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.904685185185185e-05,
+      "loss": 0.011,
+      "step": 10300
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.904592592592593e-05,
+      "loss": 0.0119,
+      "step": 10310
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.9045e-05,
+      "loss": 0.0066,
+      "step": 10320
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.904407407407408e-05,
+      "loss": 0.0099,
+      "step": 10330
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.904314814814815e-05,
+      "loss": 0.0076,
+      "step": 10340
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.904222222222222e-05,
+      "loss": 0.0073,
+      "step": 10350
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.9041296296296296e-05,
+      "loss": 0.0083,
+      "step": 10360
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.904037037037038e-05,
+      "loss": 0.0085,
+      "step": 10370
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.9039444444444446e-05,
+      "loss": 0.0088,
+      "step": 10380
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.903851851851852e-05,
+      "loss": 0.0059,
+      "step": 10390
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.9037592592592596e-05,
+      "loss": 0.0109,
+      "step": 10400
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.903666666666667e-05,
+      "loss": 0.0083,
+      "step": 10410
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.9035740740740746e-05,
+      "loss": 0.011,
+      "step": 10420
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.9034814814814814e-05,
+      "loss": 0.0077,
+      "step": 10430
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.903388888888889e-05,
+      "loss": 0.0081,
+      "step": 10440
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.9032962962962964e-05,
+      "loss": 0.009,
+      "step": 10450
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.903203703703704e-05,
+      "loss": 0.0067,
+      "step": 10460
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.903111111111111e-05,
+      "loss": 0.0067,
+      "step": 10470
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.903018518518519e-05,
+      "loss": 0.0084,
+      "step": 10480
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.9029259259259265e-05,
+      "loss": 0.0097,
+      "step": 10490
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.902833333333334e-05,
+      "loss": 0.008,
+      "step": 10500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.902740740740741e-05,
+      "loss": 0.0107,
+      "step": 10510
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.902648148148148e-05,
+      "loss": 0.0097,
+      "step": 10520
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.902555555555556e-05,
+      "loss": 0.009,
+      "step": 10530
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.902462962962963e-05,
+      "loss": 0.0072,
+      "step": 10540
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.902370370370371e-05,
+      "loss": 0.0141,
+      "step": 10550
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.9022777777777776e-05,
+      "loss": 0.0062,
+      "step": 10560
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.902185185185185e-05,
+      "loss": 0.012,
+      "step": 10570
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.9020925925925927e-05,
+      "loss": 0.0088,
+      "step": 10580
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.902e-05,
+      "loss": 0.0138,
+      "step": 10590
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.901907407407408e-05,
+      "loss": 0.0061,
+      "step": 10600
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.901814814814815e-05,
+      "loss": 0.0072,
+      "step": 10610
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.901722222222223e-05,
+      "loss": 0.0102,
+      "step": 10620
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.90162962962963e-05,
+      "loss": 0.0094,
+      "step": 10630
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.901537037037037e-05,
+      "loss": 0.0061,
+      "step": 10640
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.9014444444444445e-05,
+      "loss": 0.0108,
+      "step": 10650
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.901351851851852e-05,
+      "loss": 0.008,
+      "step": 10660
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.9012592592592595e-05,
+      "loss": 0.0081,
+      "step": 10670
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.9011666666666664e-05,
+      "loss": 0.0093,
+      "step": 10680
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.901074074074074e-05,
+      "loss": 0.0097,
+      "step": 10690
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.900981481481482e-05,
+      "loss": 0.0097,
+      "step": 10700
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.9008888888888896e-05,
+      "loss": 0.0086,
+      "step": 10710
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.9007962962962964e-05,
+      "loss": 0.0114,
+      "step": 10720
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.900703703703704e-05,
+      "loss": 0.0087,
+      "step": 10730
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.9006111111111114e-05,
+      "loss": 0.0065,
+      "step": 10740
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.900518518518519e-05,
+      "loss": 0.0061,
+      "step": 10750
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.9004259259259264e-05,
+      "loss": 0.0085,
+      "step": 10760
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.900333333333333e-05,
+      "loss": 0.009,
+      "step": 10770
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.900240740740741e-05,
+      "loss": 0.0079,
+      "step": 10780
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.900148148148148e-05,
+      "loss": 0.0079,
+      "step": 10790
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.900055555555556e-05,
+      "loss": 0.0077,
+      "step": 10800
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 0.012512334622442722,
+      "eval_runtime": 153.2052,
+      "eval_samples_per_second": 26.043,
+      "eval_steps_per_second": 3.257,
+      "step": 10800
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.899962962962963e-05,
+      "loss": 0.005,
+      "step": 10810
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.899870370370371e-05,
+      "loss": 0.009,
+      "step": 10820
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.899777777777778e-05,
+      "loss": 0.006,
+      "step": 10830
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.899685185185186e-05,
+      "loss": 0.0129,
+      "step": 10840
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.8995925925925926e-05,
+      "loss": 0.0099,
+      "step": 10850
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.8995e-05,
+      "loss": 0.0052,
+      "step": 10860
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.8994074074074076e-05,
+      "loss": 0.0084,
+      "step": 10870
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.899314814814815e-05,
+      "loss": 0.0077,
+      "step": 10880
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.899222222222222e-05,
+      "loss": 0.0071,
+      "step": 10890
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.8991296296296294e-05,
+      "loss": 0.0074,
+      "step": 10900
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.8990370370370376e-05,
+      "loss": 0.0072,
+      "step": 10910
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.898944444444445e-05,
+      "loss": 0.0108,
+      "step": 10920
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.898851851851852e-05,
+      "loss": 0.0069,
+      "step": 10930
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.8987592592592595e-05,
+      "loss": 0.0048,
+      "step": 10940
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.898666666666667e-05,
+      "loss": 0.0107,
+      "step": 10950
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.8985740740740745e-05,
+      "loss": 0.009,
+      "step": 10960
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.898481481481482e-05,
+      "loss": 0.0067,
+      "step": 10970
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.898388888888889e-05,
+      "loss": 0.0116,
+      "step": 10980
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.898296296296296e-05,
+      "loss": 0.0064,
+      "step": 10990
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.898203703703704e-05,
+      "loss": 0.0045,
+      "step": 11000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.898111111111111e-05,
+      "loss": 0.011,
+      "step": 11010
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.898018518518519e-05,
+      "loss": 0.0093,
+      "step": 11020
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8979259259259263e-05,
+      "loss": 0.0096,
+      "step": 11030
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.897833333333334e-05,
+      "loss": 0.0093,
+      "step": 11040
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8977407407407414e-05,
+      "loss": 0.0095,
+      "step": 11050
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.897648148148148e-05,
+      "loss": 0.009,
+      "step": 11060
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.897555555555556e-05,
+      "loss": 0.007,
+      "step": 11070
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.897462962962963e-05,
+      "loss": 0.009,
+      "step": 11080
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.897370370370371e-05,
+      "loss": 0.0076,
+      "step": 11090
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.8972777777777775e-05,
+      "loss": 0.009,
+      "step": 11100
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.897185185185185e-05,
+      "loss": 0.0061,
+      "step": 11110
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.8970925925925925e-05,
+      "loss": 0.0064,
+      "step": 11120
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.897000000000001e-05,
+      "loss": 0.007,
+      "step": 11130
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.8969074074074075e-05,
+      "loss": 0.0074,
+      "step": 11140
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.896814814814815e-05,
+      "loss": 0.0057,
+      "step": 11150
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.8967222222222226e-05,
+      "loss": 0.0099,
+      "step": 11160
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.89662962962963e-05,
+      "loss": 0.0083,
+      "step": 11170
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.8965370370370376e-05,
+      "loss": 0.0072,
+      "step": 11180
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.8964444444444444e-05,
+      "loss": 0.01,
+      "step": 11190
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.896351851851852e-05,
+      "loss": 0.0067,
+      "step": 11200
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.8962592592592594e-05,
+      "loss": 0.0069,
+      "step": 11210
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.896166666666667e-05,
+      "loss": 0.0063,
+      "step": 11220
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.896074074074074e-05,
+      "loss": 0.009,
+      "step": 11230
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.895981481481482e-05,
+      "loss": 0.0084,
+      "step": 11240
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.8958888888888894e-05,
+      "loss": 0.008,
+      "step": 11250
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.895796296296297e-05,
+      "loss": 0.0104,
+      "step": 11260
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.895703703703704e-05,
+      "loss": 0.0076,
+      "step": 11270
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.895611111111111e-05,
+      "loss": 0.005,
+      "step": 11280
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.895518518518519e-05,
+      "loss": 0.0102,
+      "step": 11290
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.895425925925926e-05,
+      "loss": 0.0087,
+      "step": 11300
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.895333333333333e-05,
+      "loss": 0.0079,
+      "step": 11310
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.8952407407407406e-05,
+      "loss": 0.0106,
+      "step": 11320
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.895148148148148e-05,
+      "loss": 0.008,
+      "step": 11330
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.895055555555556e-05,
+      "loss": 0.0093,
+      "step": 11340
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.894962962962963e-05,
+      "loss": 0.009,
+      "step": 11350
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.8948703703703706e-05,
+      "loss": 0.0085,
+      "step": 11360
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.894777777777778e-05,
+      "loss": 0.0092,
+      "step": 11370
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.8946851851851856e-05,
+      "loss": 0.0108,
+      "step": 11380
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.894592592592593e-05,
+      "loss": 0.0071,
+      "step": 11390
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.8945e-05,
+      "loss": 0.0073,
+      "step": 11400
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.8944074074074075e-05,
+      "loss": 0.0082,
+      "step": 11410
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.894314814814815e-05,
+      "loss": 0.006,
+      "step": 11420
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.8942222222222225e-05,
+      "loss": 0.0127,
+      "step": 11430
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.894129629629629e-05,
+      "loss": 0.0114,
+      "step": 11440
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.8940370370370375e-05,
+      "loss": 0.0058,
+      "step": 11450
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.893944444444445e-05,
+      "loss": 0.0078,
+      "step": 11460
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.8938518518518525e-05,
+      "loss": 0.0067,
+      "step": 11470
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.8937592592592593e-05,
+      "loss": 0.0104,
+      "step": 11480
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.893666666666667e-05,
+      "loss": 0.0072,
+      "step": 11490
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.8935740740740744e-05,
+      "loss": 0.0065,
+      "step": 11500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.893481481481482e-05,
+      "loss": 0.0059,
+      "step": 11510
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.893388888888889e-05,
+      "loss": 0.0071,
+      "step": 11520
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.893296296296296e-05,
+      "loss": 0.0077,
+      "step": 11530
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.893203703703704e-05,
+      "loss": 0.0068,
+      "step": 11540
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.893111111111111e-05,
+      "loss": 0.0071,
+      "step": 11550
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.893018518518519e-05,
+      "loss": 0.0087,
+      "step": 11560
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.892925925925926e-05,
+      "loss": 0.0064,
+      "step": 11570
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.892833333333334e-05,
+      "loss": 0.008,
+      "step": 11580
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.892740740740741e-05,
+      "loss": 0.0073,
+      "step": 11590
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.892648148148149e-05,
+      "loss": 0.0096,
+      "step": 11600
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.8925555555555556e-05,
+      "loss": 0.0109,
+      "step": 11610
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.892462962962963e-05,
+      "loss": 0.0091,
+      "step": 11620
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.8923703703703706e-05,
+      "loss": 0.0084,
+      "step": 11630
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.892277777777778e-05,
+      "loss": 0.0091,
+      "step": 11640
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.892185185185185e-05,
+      "loss": 0.0077,
+      "step": 11650
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.8920925925925924e-05,
+      "loss": 0.0094,
+      "step": 11660
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.8920000000000006e-05,
+      "loss": 0.0092,
+      "step": 11670
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.891907407407408e-05,
+      "loss": 0.0055,
+      "step": 11680
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.891814814814815e-05,
+      "loss": 0.0044,
+      "step": 11690
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.8917222222222224e-05,
+      "loss": 0.0049,
+      "step": 11700
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.89162962962963e-05,
+      "loss": 0.0078,
+      "step": 11710
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.8915370370370374e-05,
+      "loss": 0.0062,
+      "step": 11720
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.891444444444444e-05,
+      "loss": 0.0105,
+      "step": 11730
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.891351851851852e-05,
+      "loss": 0.0082,
+      "step": 11740
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.891259259259259e-05,
+      "loss": 0.0153,
+      "step": 11750
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.891166666666667e-05,
+      "loss": 0.0089,
+      "step": 11760
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.891074074074074e-05,
+      "loss": 0.0105,
+      "step": 11770
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.890981481481482e-05,
+      "loss": 0.01,
+      "step": 11780
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.890888888888889e-05,
+      "loss": 0.0051,
+      "step": 11790
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.890796296296297e-05,
+      "loss": 0.0075,
+      "step": 11800
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.890703703703704e-05,
+      "loss": 0.0095,
+      "step": 11810
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.890611111111111e-05,
+      "loss": 0.0058,
+      "step": 11820
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.8905185185185187e-05,
+      "loss": 0.0105,
+      "step": 11830
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.890425925925926e-05,
+      "loss": 0.0048,
+      "step": 11840
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.890333333333334e-05,
+      "loss": 0.0071,
+      "step": 11850
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.8902407407407405e-05,
+      "loss": 0.0083,
+      "step": 11860
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.890148148148148e-05,
+      "loss": 0.0071,
+      "step": 11870
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.890055555555556e-05,
+      "loss": 0.0114,
+      "step": 11880
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.889962962962964e-05,
+      "loss": 0.0088,
+      "step": 11890
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.8898703703703705e-05,
+      "loss": 0.0102,
+      "step": 11900
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.889777777777778e-05,
+      "loss": 0.0074,
+      "step": 11910
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.8896851851851855e-05,
+      "loss": 0.0114,
+      "step": 11920
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.889592592592593e-05,
+      "loss": 0.0082,
+      "step": 11930
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.8895e-05,
+      "loss": 0.008,
+      "step": 11940
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.8894074074074074e-05,
+      "loss": 0.0046,
+      "step": 11950
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.889314814814815e-05,
+      "loss": 0.0086,
+      "step": 11960
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.8892222222222224e-05,
+      "loss": 0.0095,
+      "step": 11970
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.88912962962963e-05,
+      "loss": 0.0077,
+      "step": 11980
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.8890370370370374e-05,
+      "loss": 0.0082,
+      "step": 11990
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.888944444444445e-05,
+      "loss": 0.008,
+      "step": 12000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.8888518518518524e-05,
+      "loss": 0.0096,
+      "step": 12010
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.88875925925926e-05,
+      "loss": 0.0083,
+      "step": 12020
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.888666666666667e-05,
+      "loss": 0.0089,
+      "step": 12030
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.888574074074074e-05,
+      "loss": 0.0063,
+      "step": 12040
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.888481481481482e-05,
+      "loss": 0.0073,
+      "step": 12050
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.888388888888889e-05,
+      "loss": 0.0087,
+      "step": 12060
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.888296296296296e-05,
+      "loss": 0.0058,
+      "step": 12070
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.8882037037037036e-05,
+      "loss": 0.0108,
+      "step": 12080
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.888111111111111e-05,
+      "loss": 0.0074,
+      "step": 12090
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.888018518518519e-05,
+      "loss": 0.0101,
+      "step": 12100
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.887925925925926e-05,
+      "loss": 0.0072,
+      "step": 12110
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.8878333333333336e-05,
+      "loss": 0.0122,
+      "step": 12120
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.887740740740741e-05,
+      "loss": 0.0067,
+      "step": 12130
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.8876481481481486e-05,
+      "loss": 0.0062,
+      "step": 12140
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.8875555555555554e-05,
+      "loss": 0.0088,
+      "step": 12150
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.887462962962963e-05,
+      "loss": 0.0049,
+      "step": 12160
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.8873703703703705e-05,
+      "loss": 0.0088,
+      "step": 12170
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.887277777777778e-05,
+      "loss": 0.0092,
+      "step": 12180
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.8871851851851855e-05,
+      "loss": 0.0042,
+      "step": 12190
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.887092592592592e-05,
+      "loss": 0.0086,
+      "step": 12200
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.8870000000000005e-05,
+      "loss": 0.0061,
+      "step": 12210
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.886907407407408e-05,
+      "loss": 0.0087,
+      "step": 12220
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.8868148148148155e-05,
+      "loss": 0.0069,
+      "step": 12230
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.886722222222222e-05,
+      "loss": 0.01,
+      "step": 12240
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.88662962962963e-05,
+      "loss": 0.0064,
+      "step": 12250
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.886537037037037e-05,
+      "loss": 0.0122,
+      "step": 12260
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.886444444444445e-05,
+      "loss": 0.0061,
+      "step": 12270
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.8863518518518517e-05,
+      "loss": 0.0095,
+      "step": 12280
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.886259259259259e-05,
+      "loss": 0.0078,
+      "step": 12290
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.886166666666667e-05,
+      "loss": 0.0069,
+      "step": 12300
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.886074074074075e-05,
+      "loss": 0.0064,
+      "step": 12310
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.885981481481482e-05,
+      "loss": 0.0063,
+      "step": 12320
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.885888888888889e-05,
+      "loss": 0.0072,
+      "step": 12330
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.885796296296297e-05,
+      "loss": 0.0057,
+      "step": 12340
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.885703703703704e-05,
+      "loss": 0.0073,
+      "step": 12350
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.885611111111111e-05,
+      "loss": 0.0103,
+      "step": 12360
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.8855185185185185e-05,
+      "loss": 0.0075,
+      "step": 12370
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.885435185185186e-05,
+      "loss": 0.0093,
+      "step": 12380
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.8853425925925925e-05,
+      "loss": 0.0093,
+      "step": 12390
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.88525e-05,
+      "loss": 0.0093,
+      "step": 12400
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.8851574074074075e-05,
+      "loss": 0.0087,
+      "step": 12410
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.885064814814815e-05,
+      "loss": 0.0096,
+      "step": 12420
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.8849722222222225e-05,
+      "loss": 0.0059,
+      "step": 12430
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.88487962962963e-05,
+      "loss": 0.0055,
+      "step": 12440
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.8847870370370376e-05,
+      "loss": 0.009,
+      "step": 12450
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.884694444444445e-05,
+      "loss": 0.0062,
+      "step": 12460
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.884601851851852e-05,
+      "loss": 0.0091,
+      "step": 12470
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.8845092592592594e-05,
+      "loss": 0.01,
+      "step": 12480
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.884416666666667e-05,
+      "loss": 0.0066,
+      "step": 12490
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.8843240740740744e-05,
+      "loss": 0.0089,
+      "step": 12500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.884231481481482e-05,
+      "loss": 0.0083,
+      "step": 12510
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.884138888888889e-05,
+      "loss": 0.0105,
+      "step": 12520
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.884046296296296e-05,
+      "loss": 0.0081,
+      "step": 12530
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.8839537037037044e-05,
+      "loss": 0.0076,
+      "step": 12540
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.883861111111111e-05,
+      "loss": 0.0111,
+      "step": 12550
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.883768518518519e-05,
+      "loss": 0.0087,
+      "step": 12560
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.883675925925926e-05,
+      "loss": 0.0117,
+      "step": 12570
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.883583333333334e-05,
+      "loss": 0.0072,
+      "step": 12580
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.883490740740741e-05,
+      "loss": 0.0073,
+      "step": 12590
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.883398148148148e-05,
+      "loss": 0.0073,
+      "step": 12600
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.8833055555555556e-05,
+      "loss": 0.0054,
+      "step": 12610
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.883212962962963e-05,
+      "loss": 0.0046,
+      "step": 12620
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.8831203703703706e-05,
+      "loss": 0.0085,
+      "step": 12630
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.8830277777777774e-05,
+      "loss": 0.0077,
+      "step": 12640
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.8829351851851856e-05,
+      "loss": 0.0123,
+      "step": 12650
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.882842592592593e-05,
+      "loss": 0.0108,
+      "step": 12660
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.8827500000000006e-05,
+      "loss": 0.0075,
+      "step": 12670
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.8826574074074075e-05,
+      "loss": 0.0081,
+      "step": 12680
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.882564814814815e-05,
+      "loss": 0.0069,
+      "step": 12690
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.8824722222222225e-05,
+      "loss": 0.0059,
+      "step": 12700
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.88237962962963e-05,
+      "loss": 0.0069,
+      "step": 12710
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.8822870370370375e-05,
+      "loss": 0.0087,
+      "step": 12720
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.882194444444444e-05,
+      "loss": 0.0064,
+      "step": 12730
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.882101851851852e-05,
+      "loss": 0.0111,
+      "step": 12740
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.882009259259259e-05,
+      "loss": 0.0093,
+      "step": 12750
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.881916666666667e-05,
+      "loss": 0.0064,
+      "step": 12760
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.8818240740740743e-05,
+      "loss": 0.0073,
+      "step": 12770
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.881731481481482e-05,
+      "loss": 0.0084,
+      "step": 12780
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.8816388888888894e-05,
+      "loss": 0.0071,
+      "step": 12790
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.881546296296297e-05,
+      "loss": 0.0068,
+      "step": 12800
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.881453703703704e-05,
+      "loss": 0.0054,
+      "step": 12810
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.881361111111111e-05,
+      "loss": 0.0063,
+      "step": 12820
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.881268518518519e-05,
+      "loss": 0.0058,
+      "step": 12830
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.881175925925926e-05,
+      "loss": 0.008,
+      "step": 12840
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.881083333333333e-05,
+      "loss": 0.0064,
+      "step": 12850
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.8809907407407405e-05,
+      "loss": 0.0052,
+      "step": 12860
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.880898148148149e-05,
+      "loss": 0.0102,
+      "step": 12870
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.880805555555556e-05,
+      "loss": 0.0095,
+      "step": 12880
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.880712962962963e-05,
+      "loss": 0.0098,
+      "step": 12890
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.8806203703703706e-05,
+      "loss": 0.0068,
+      "step": 12900
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.880527777777778e-05,
+      "loss": 0.008,
+      "step": 12910
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.8804351851851856e-05,
+      "loss": 0.0088,
+      "step": 12920
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.880342592592593e-05,
+      "loss": 0.0072,
+      "step": 12930
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.88025e-05,
+      "loss": 0.0067,
+      "step": 12940
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.8801574074074074e-05,
+      "loss": 0.007,
+      "step": 12950
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.880064814814815e-05,
+      "loss": 0.0104,
+      "step": 12960
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.8799722222222224e-05,
+      "loss": 0.0095,
+      "step": 12970
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.87987962962963e-05,
+      "loss": 0.0058,
+      "step": 12980
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.8797870370370374e-05,
+      "loss": 0.0086,
+      "step": 12990
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.879694444444445e-05,
+      "loss": 0.0083,
+      "step": 13000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.8796018518518524e-05,
+      "loss": 0.0069,
+      "step": 13010
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.879509259259259e-05,
+      "loss": 0.0096,
+      "step": 13020
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.879416666666667e-05,
+      "loss": 0.0075,
+      "step": 13030
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.879324074074074e-05,
+      "loss": 0.009,
+      "step": 13040
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.879231481481482e-05,
+      "loss": 0.0062,
+      "step": 13050
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.8791388888888886e-05,
+      "loss": 0.004,
+      "step": 13060
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.879046296296296e-05,
+      "loss": 0.0095,
+      "step": 13070
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.878953703703704e-05,
+      "loss": 0.0049,
+      "step": 13080
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.878861111111112e-05,
+      "loss": 0.0066,
+      "step": 13090
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.8787685185185186e-05,
+      "loss": 0.0061,
+      "step": 13100
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.878675925925926e-05,
+      "loss": 0.0077,
+      "step": 13110
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.8785833333333336e-05,
+      "loss": 0.0069,
+      "step": 13120
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.878490740740741e-05,
+      "loss": 0.0086,
+      "step": 13130
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.8783981481481487e-05,
+      "loss": 0.0065,
+      "step": 13140
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.8783055555555555e-05,
+      "loss": 0.0086,
+      "step": 13150
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.878212962962963e-05,
+      "loss": 0.0071,
+      "step": 13160
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.8781203703703705e-05,
+      "loss": 0.0088,
+      "step": 13170
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.878027777777778e-05,
+      "loss": 0.011,
+      "step": 13180
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.8779351851851855e-05,
+      "loss": 0.0099,
+      "step": 13190
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.877842592592593e-05,
+      "loss": 0.0079,
+      "step": 13200
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.8777500000000005e-05,
+      "loss": 0.003,
+      "step": 13210
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.877657407407408e-05,
+      "loss": 0.0054,
+      "step": 13220
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.877564814814815e-05,
+      "loss": 0.0063,
+      "step": 13230
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.8774722222222224e-05,
+      "loss": 0.0095,
+      "step": 13240
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.87737962962963e-05,
+      "loss": 0.0065,
+      "step": 13250
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.8772870370370374e-05,
+      "loss": 0.006,
+      "step": 13260
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.877194444444444e-05,
+      "loss": 0.008,
+      "step": 13270
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.877101851851852e-05,
+      "loss": 0.0096,
+      "step": 13280
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.877009259259259e-05,
+      "loss": 0.0079,
+      "step": 13290
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.8769166666666674e-05,
+      "loss": 0.0102,
+      "step": 13300
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.876824074074074e-05,
+      "loss": 0.0066,
+      "step": 13310
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.876731481481482e-05,
+      "loss": 0.0051,
+      "step": 13320
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.876638888888889e-05,
+      "loss": 0.0066,
+      "step": 13330
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.876546296296297e-05,
+      "loss": 0.0107,
+      "step": 13340
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.876453703703704e-05,
+      "loss": 0.0085,
+      "step": 13350
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.876361111111111e-05,
+      "loss": 0.0092,
+      "step": 13360
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.8762685185185186e-05,
+      "loss": 0.0081,
+      "step": 13370
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.876175925925926e-05,
+      "loss": 0.0094,
+      "step": 13380
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.8760833333333336e-05,
+      "loss": 0.0091,
+      "step": 13390
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.875990740740741e-05,
+      "loss": 0.0085,
+      "step": 13400
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.8758981481481486e-05,
+      "loss": 0.0072,
+      "step": 13410
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.875805555555556e-05,
+      "loss": 0.0071,
+      "step": 13420
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.8757129629629636e-05,
+      "loss": 0.0081,
+      "step": 13430
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.8756203703703704e-05,
+      "loss": 0.0054,
+      "step": 13440
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.875527777777778e-05,
+      "loss": 0.0104,
+      "step": 13450
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.8754351851851854e-05,
+      "loss": 0.0076,
+      "step": 13460
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.875342592592593e-05,
+      "loss": 0.0087,
+      "step": 13470
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.87525e-05,
+      "loss": 0.0072,
+      "step": 13480
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.875157407407407e-05,
+      "loss": 0.0103,
+      "step": 13490
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.875064814814815e-05,
+      "loss": 0.0071,
+      "step": 13500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.874972222222223e-05,
+      "loss": 0.0087,
+      "step": 13510
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.87487962962963e-05,
+      "loss": 0.0058,
+      "step": 13520
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.874787037037037e-05,
+      "loss": 0.0044,
+      "step": 13530
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.874694444444445e-05,
+      "loss": 0.0054,
+      "step": 13540
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.874601851851852e-05,
+      "loss": 0.0113,
+      "step": 13550
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.87450925925926e-05,
+      "loss": 0.0065,
+      "step": 13560
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.8744166666666667e-05,
+      "loss": 0.0086,
+      "step": 13570
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.874324074074074e-05,
+      "loss": 0.0085,
+      "step": 13580
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.874231481481482e-05,
+      "loss": 0.0064,
+      "step": 13590
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.874138888888889e-05,
+      "loss": 0.0067,
+      "step": 13600
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.874046296296296e-05,
+      "loss": 0.0049,
+      "step": 13610
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.873953703703704e-05,
+      "loss": 0.0093,
+      "step": 13620
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.873861111111112e-05,
+      "loss": 0.008,
+      "step": 13630
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.873768518518519e-05,
+      "loss": 0.0099,
+      "step": 13640
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.873675925925926e-05,
+      "loss": 0.006,
+      "step": 13650
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.8735833333333335e-05,
+      "loss": 0.009,
+      "step": 13660
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.873490740740741e-05,
+      "loss": 0.0057,
+      "step": 13670
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.8733981481481485e-05,
+      "loss": 0.0101,
+      "step": 13680
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.8733055555555554e-05,
+      "loss": 0.0056,
+      "step": 13690
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.873212962962963e-05,
+      "loss": 0.007,
+      "step": 13700
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.8731203703703704e-05,
+      "loss": 0.0044,
+      "step": 13710
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.873027777777778e-05,
+      "loss": 0.0085,
+      "step": 13720
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.8729351851851854e-05,
+      "loss": 0.0062,
+      "step": 13730
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.872842592592593e-05,
+      "loss": 0.0049,
+      "step": 13740
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.8727500000000004e-05,
+      "loss": 0.0082,
+      "step": 13750
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.872657407407408e-05,
+      "loss": 0.0091,
+      "step": 13760
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.8725648148148154e-05,
+      "loss": 0.0065,
+      "step": 13770
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.872472222222222e-05,
+      "loss": 0.0066,
+      "step": 13780
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.87237962962963e-05,
+      "loss": 0.0081,
+      "step": 13790
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.872287037037037e-05,
+      "loss": 0.0084,
+      "step": 13800
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.872194444444445e-05,
+      "loss": 0.0099,
+      "step": 13810
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.8721018518518516e-05,
+      "loss": 0.0114,
+      "step": 13820
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.872009259259259e-05,
+      "loss": 0.0079,
+      "step": 13830
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.871916666666667e-05,
+      "loss": 0.0069,
+      "step": 13840
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.871824074074075e-05,
+      "loss": 0.0091,
+      "step": 13850
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.8717314814814816e-05,
+      "loss": 0.005,
+      "step": 13860
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.871638888888889e-05,
+      "loss": 0.0072,
+      "step": 13870
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.8715462962962966e-05,
+      "loss": 0.0076,
+      "step": 13880
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.871453703703704e-05,
+      "loss": 0.004,
+      "step": 13890
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.871361111111111e-05,
+      "loss": 0.0085,
+      "step": 13900
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.8712685185185185e-05,
+      "loss": 0.0083,
+      "step": 13910
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.871175925925926e-05,
+      "loss": 0.0096,
+      "step": 13920
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.8710833333333335e-05,
+      "loss": 0.0089,
+      "step": 13930
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.870990740740741e-05,
+      "loss": 0.007,
+      "step": 13940
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.8708981481481485e-05,
+      "loss": 0.005,
+      "step": 13950
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.870805555555556e-05,
+      "loss": 0.0059,
+      "step": 13960
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.8707129629629635e-05,
+      "loss": 0.0067,
+      "step": 13970
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.870620370370371e-05,
+      "loss": 0.0071,
+      "step": 13980
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.870527777777778e-05,
+      "loss": 0.0056,
+      "step": 13990
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.870435185185185e-05,
+      "loss": 0.0076,
+      "step": 14000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.870342592592593e-05,
+      "loss": 0.0077,
+      "step": 14010
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.87025e-05,
+      "loss": 0.0071,
+      "step": 14020
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.870157407407407e-05,
+      "loss": 0.0074,
+      "step": 14030
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.870064814814815e-05,
+      "loss": 0.009,
+      "step": 14040
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.869972222222223e-05,
+      "loss": 0.0099,
+      "step": 14050
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.8698796296296304e-05,
+      "loss": 0.0084,
+      "step": 14060
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.869787037037037e-05,
+      "loss": 0.0065,
+      "step": 14070
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.869694444444445e-05,
+      "loss": 0.0062,
+      "step": 14080
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.869601851851852e-05,
+      "loss": 0.0088,
+      "step": 14090
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.86950925925926e-05,
+      "loss": 0.0058,
+      "step": 14100
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.8694166666666665e-05,
+      "loss": 0.0041,
+      "step": 14110
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.869324074074074e-05,
+      "loss": 0.0058,
+      "step": 14120
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.8692314814814815e-05,
+      "loss": 0.0095,
+      "step": 14130
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.869138888888889e-05,
+      "loss": 0.0122,
+      "step": 14140
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.8690462962962966e-05,
+      "loss": 0.0055,
+      "step": 14150
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.868953703703704e-05,
+      "loss": 0.0087,
+      "step": 14160
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.8688611111111116e-05,
+      "loss": 0.0078,
+      "step": 14170
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.868768518518519e-05,
+      "loss": 0.0076,
+      "step": 14180
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.8686759259259266e-05,
+      "loss": 0.007,
+      "step": 14190
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.8685833333333334e-05,
+      "loss": 0.0053,
+      "step": 14200
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.868490740740741e-05,
+      "loss": 0.0066,
+      "step": 14210
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.8683981481481484e-05,
+      "loss": 0.005,
+      "step": 14220
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.868305555555556e-05,
+      "loss": 0.0081,
+      "step": 14230
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.868212962962963e-05,
+      "loss": 0.0073,
+      "step": 14240
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.86812037037037e-05,
+      "loss": 0.0072,
+      "step": 14250
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.868027777777778e-05,
+      "loss": 0.0089,
+      "step": 14260
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.867935185185186e-05,
+      "loss": 0.0105,
+      "step": 14270
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.867842592592593e-05,
+      "loss": 0.0059,
+      "step": 14280
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.86775e-05,
+      "loss": 0.0069,
+      "step": 14290
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.867657407407408e-05,
+      "loss": 0.0079,
+      "step": 14300
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.867564814814815e-05,
+      "loss": 0.0065,
+      "step": 14310
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.867472222222222e-05,
+      "loss": 0.0102,
+      "step": 14320
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.8673796296296296e-05,
+      "loss": 0.0074,
+      "step": 14330
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.867287037037037e-05,
+      "loss": 0.0057,
+      "step": 14340
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.8671944444444446e-05,
+      "loss": 0.0057,
+      "step": 14350
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.867101851851852e-05,
+      "loss": 0.0053,
+      "step": 14360
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.8670092592592596e-05,
+      "loss": 0.0053,
+      "step": 14370
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.866916666666667e-05,
+      "loss": 0.0076,
+      "step": 14380
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.8668240740740747e-05,
+      "loss": 0.0088,
+      "step": 14390
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.866731481481482e-05,
+      "loss": 0.0105,
+      "step": 14400
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.866638888888889e-05,
+      "loss": 0.0059,
+      "step": 14410
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.8665462962962965e-05,
+      "loss": 0.008,
+      "step": 14420
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.866453703703704e-05,
+      "loss": 0.0076,
+      "step": 14430
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.8663611111111115e-05,
+      "loss": 0.008,
+      "step": 14440
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.866268518518518e-05,
+      "loss": 0.0092,
+      "step": 14450
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.866175925925926e-05,
+      "loss": 0.0078,
+      "step": 14460
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.8660833333333333e-05,
+      "loss": 0.0107,
+      "step": 14470
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.8659907407407415e-05,
+      "loss": 0.0081,
+      "step": 14480
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.8658981481481484e-05,
+      "loss": 0.0079,
+      "step": 14490
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.865805555555556e-05,
+      "loss": 0.0086,
+      "step": 14500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.8657129629629634e-05,
+      "loss": 0.0081,
+      "step": 14510
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.865620370370371e-05,
+      "loss": 0.0061,
+      "step": 14520
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.865527777777778e-05,
+      "loss": 0.0068,
+      "step": 14530
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.865435185185185e-05,
+      "loss": 0.0093,
+      "step": 14540
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.865342592592593e-05,
+      "loss": 0.0057,
+      "step": 14550
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.86525e-05,
+      "loss": 0.0087,
+      "step": 14560
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.865157407407408e-05,
+      "loss": 0.0051,
+      "step": 14570
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.8650648148148146e-05,
+      "loss": 0.006,
+      "step": 14580
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.864972222222223e-05,
+      "loss": 0.0071,
+      "step": 14590
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.86487962962963e-05,
+      "loss": 0.0081,
+      "step": 14600
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.864796296296297e-05,
+      "loss": 0.0102,
+      "step": 14610
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.864703703703704e-05,
+      "loss": 0.0104,
+      "step": 14620
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.864611111111112e-05,
+      "loss": 0.0095,
+      "step": 14630
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.8645185185185186e-05,
+      "loss": 0.0072,
+      "step": 14640
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.864425925925926e-05,
+      "loss": 0.008,
+      "step": 14650
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.8643333333333336e-05,
+      "loss": 0.006,
+      "step": 14660
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.864240740740741e-05,
+      "loss": 0.0089,
+      "step": 14670
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.864148148148148e-05,
+      "loss": 0.0054,
+      "step": 14680
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.8640555555555554e-05,
+      "loss": 0.0046,
+      "step": 14690
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.863962962962963e-05,
+      "loss": 0.0086,
+      "step": 14700
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.863870370370371e-05,
+      "loss": 0.0077,
+      "step": 14710
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.863777777777778e-05,
+      "loss": 0.007,
+      "step": 14720
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.8636851851851854e-05,
+      "loss": 0.0073,
+      "step": 14730
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.863592592592593e-05,
+      "loss": 0.0056,
+      "step": 14740
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.8635000000000004e-05,
+      "loss": 0.0092,
+      "step": 14750
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.863407407407408e-05,
+      "loss": 0.0059,
+      "step": 14760
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.863314814814815e-05,
+      "loss": 0.0092,
+      "step": 14770
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.863222222222222e-05,
+      "loss": 0.0075,
+      "step": 14780
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.86312962962963e-05,
+      "loss": 0.0066,
+      "step": 14790
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.863037037037037e-05,
+      "loss": 0.0069,
+      "step": 14800
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.862944444444444e-05,
+      "loss": 0.0043,
+      "step": 14810
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.862851851851852e-05,
+      "loss": 0.0112,
+      "step": 14820
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.86275925925926e-05,
+      "loss": 0.0063,
+      "step": 14830
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.862666666666667e-05,
+      "loss": 0.0051,
+      "step": 14840
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.862574074074074e-05,
+      "loss": 0.0116,
+      "step": 14850
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.8624814814814817e-05,
+      "loss": 0.0073,
+      "step": 14860
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.862388888888889e-05,
+      "loss": 0.0048,
+      "step": 14870
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.862296296296297e-05,
+      "loss": 0.0067,
+      "step": 14880
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.8622037037037035e-05,
+      "loss": 0.0059,
+      "step": 14890
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.862111111111111e-05,
+      "loss": 0.0063,
+      "step": 14900
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.8620185185185185e-05,
+      "loss": 0.0085,
+      "step": 14910
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.861925925925926e-05,
+      "loss": 0.0079,
+      "step": 14920
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.8618333333333335e-05,
+      "loss": 0.01,
+      "step": 14930
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.861740740740741e-05,
+      "loss": 0.0069,
+      "step": 14940
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.8616481481481485e-05,
+      "loss": 0.0083,
+      "step": 14950
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.861555555555556e-05,
+      "loss": 0.008,
+      "step": 14960
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.8614629629629635e-05,
+      "loss": 0.009,
+      "step": 14970
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.8613703703703704e-05,
+      "loss": 0.0134,
+      "step": 14980
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.861277777777778e-05,
+      "loss": 0.0066,
+      "step": 14990
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.8611851851851854e-05,
+      "loss": 0.0056,
+      "step": 15000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.861092592592593e-05,
+      "loss": 0.0083,
+      "step": 15010
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.861e-05,
+      "loss": 0.0069,
+      "step": 15020
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.860907407407408e-05,
+      "loss": 0.0041,
+      "step": 15030
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.8608148148148154e-05,
+      "loss": 0.0043,
+      "step": 15040
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.860722222222223e-05,
+      "loss": 0.0085,
+      "step": 15050
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.86062962962963e-05,
+      "loss": 0.0078,
+      "step": 15060
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.860537037037037e-05,
+      "loss": 0.0092,
+      "step": 15070
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.860444444444445e-05,
+      "loss": 0.0044,
+      "step": 15080
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.860351851851852e-05,
+      "loss": 0.0083,
+      "step": 15090
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.860259259259259e-05,
+      "loss": 0.0084,
+      "step": 15100
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.8601666666666666e-05,
+      "loss": 0.0071,
+      "step": 15110
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.860074074074074e-05,
+      "loss": 0.0037,
+      "step": 15120
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.8599814814814816e-05,
+      "loss": 0.0061,
+      "step": 15130
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.859888888888889e-05,
+      "loss": 0.0049,
+      "step": 15140
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.8597962962962966e-05,
+      "loss": 0.0103,
+      "step": 15150
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.859703703703704e-05,
+      "loss": 0.0098,
+      "step": 15160
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.8596111111111116e-05,
+      "loss": 0.0088,
+      "step": 15170
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.859518518518519e-05,
+      "loss": 0.0056,
+      "step": 15180
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.859425925925926e-05,
+      "loss": 0.0075,
+      "step": 15190
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.8593333333333335e-05,
+      "loss": 0.0056,
+      "step": 15200
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.859240740740741e-05,
+      "loss": 0.0037,
+      "step": 15210
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.8591481481481485e-05,
+      "loss": 0.0092,
+      "step": 15220
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.859055555555555e-05,
+      "loss": 0.0115,
+      "step": 15230
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.858962962962963e-05,
+      "loss": 0.0118,
+      "step": 15240
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.858870370370371e-05,
+      "loss": 0.007,
+      "step": 15250
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.8587777777777785e-05,
+      "loss": 0.0069,
+      "step": 15260
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.858685185185185e-05,
+      "loss": 0.0075,
+      "step": 15270
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.858592592592593e-05,
+      "loss": 0.0088,
+      "step": 15280
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.8585e-05,
+      "loss": 0.0086,
+      "step": 15290
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.858407407407408e-05,
+      "loss": 0.013,
+      "step": 15300
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.8583148148148147e-05,
+      "loss": 0.0059,
+      "step": 15310
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.858222222222222e-05,
+      "loss": 0.0074,
+      "step": 15320
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.85812962962963e-05,
+      "loss": 0.0089,
+      "step": 15330
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.858037037037037e-05,
+      "loss": 0.0058,
+      "step": 15340
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.857944444444445e-05,
+      "loss": 0.0066,
+      "step": 15350
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.857851851851852e-05,
+      "loss": 0.0096,
+      "step": 15360
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.85775925925926e-05,
+      "loss": 0.0046,
+      "step": 15370
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.857666666666667e-05,
+      "loss": 0.0043,
+      "step": 15380
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.857574074074075e-05,
+      "loss": 0.0078,
+      "step": 15390
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.8574814814814815e-05,
+      "loss": 0.0054,
+      "step": 15400
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.857388888888889e-05,
+      "loss": 0.0054,
+      "step": 15410
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.8572962962962965e-05,
+      "loss": 0.0085,
+      "step": 15420
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.857203703703704e-05,
+      "loss": 0.0078,
+      "step": 15430
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.857111111111111e-05,
+      "loss": 0.0075,
+      "step": 15440
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.8570185185185184e-05,
+      "loss": 0.0057,
+      "step": 15450
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.8569259259259266e-05,
+      "loss": 0.0064,
+      "step": 15460
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.856833333333334e-05,
+      "loss": 0.0073,
+      "step": 15470
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.856740740740741e-05,
+      "loss": 0.0096,
+      "step": 15480
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.8566481481481484e-05,
+      "loss": 0.0071,
+      "step": 15490
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.856555555555556e-05,
+      "loss": 0.0074,
+      "step": 15500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.8564629629629634e-05,
+      "loss": 0.0047,
+      "step": 15510
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.85637037037037e-05,
+      "loss": 0.0045,
+      "step": 15520
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.856277777777778e-05,
+      "loss": 0.0074,
+      "step": 15530
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.856185185185185e-05,
+      "loss": 0.0066,
+      "step": 15540
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.856092592592593e-05,
+      "loss": 0.0054,
+      "step": 15550
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.856e-05,
+      "loss": 0.0068,
+      "step": 15560
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.855907407407408e-05,
+      "loss": 0.0068,
+      "step": 15570
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.855814814814815e-05,
+      "loss": 0.0075,
+      "step": 15580
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.855722222222223e-05,
+      "loss": 0.0069,
+      "step": 15590
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.85562962962963e-05,
+      "loss": 0.0101,
+      "step": 15600
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.855537037037037e-05,
+      "loss": 0.0086,
+      "step": 15610
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.8554444444444446e-05,
+      "loss": 0.0108,
+      "step": 15620
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.855351851851852e-05,
+      "loss": 0.0058,
+      "step": 15630
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.8552592592592596e-05,
+      "loss": 0.0112,
+      "step": 15640
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.8551666666666665e-05,
+      "loss": 0.0078,
+      "step": 15650
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.855074074074074e-05,
+      "loss": 0.008,
+      "step": 15660
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.8549814814814815e-05,
+      "loss": 0.0095,
+      "step": 15670
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.8548888888888897e-05,
+      "loss": 0.0104,
+      "step": 15680
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8547962962962965e-05,
+      "loss": 0.0073,
+      "step": 15690
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.854703703703704e-05,
+      "loss": 0.0149,
+      "step": 15700
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8546111111111115e-05,
+      "loss": 0.0059,
+      "step": 15710
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.854518518518519e-05,
+      "loss": 0.0092,
+      "step": 15720
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.854425925925926e-05,
+      "loss": 0.0118,
+      "step": 15730
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.854333333333333e-05,
+      "loss": 0.0057,
+      "step": 15740
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.854240740740741e-05,
+      "loss": 0.0085,
+      "step": 15750
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.8541481481481483e-05,
+      "loss": 0.0063,
+      "step": 15760
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.854055555555556e-05,
+      "loss": 0.0055,
+      "step": 15770
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.853962962962963e-05,
+      "loss": 0.0067,
+      "step": 15780
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.853870370370371e-05,
+      "loss": 0.0069,
+      "step": 15790
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.8537777777777784e-05,
+      "loss": 0.0065,
+      "step": 15800
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.853685185185186e-05,
+      "loss": 0.0091,
+      "step": 15810
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.853592592592593e-05,
+      "loss": 0.0073,
+      "step": 15820
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.8535e-05,
+      "loss": 0.0087,
+      "step": 15830
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.853416666666667e-05,
+      "loss": 0.0075,
+      "step": 15840
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.853324074074074e-05,
+      "loss": 0.0044,
+      "step": 15850
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.853231481481482e-05,
+      "loss": 0.0073,
+      "step": 15860
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.853138888888889e-05,
+      "loss": 0.0075,
+      "step": 15870
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.853046296296296e-05,
+      "loss": 0.0085,
+      "step": 15880
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.8529537037037035e-05,
+      "loss": 0.0074,
+      "step": 15890
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.852861111111111e-05,
+      "loss": 0.0075,
+      "step": 15900
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.852768518518519e-05,
+      "loss": 0.008,
+      "step": 15910
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.852675925925926e-05,
+      "loss": 0.0082,
+      "step": 15920
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.8525833333333336e-05,
+      "loss": 0.0051,
+      "step": 15930
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.852490740740741e-05,
+      "loss": 0.0092,
+      "step": 15940
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.8523981481481486e-05,
+      "loss": 0.0073,
+      "step": 15950
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.852305555555556e-05,
+      "loss": 0.0049,
+      "step": 15960
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.852212962962963e-05,
+      "loss": 0.0068,
+      "step": 15970
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.8521203703703704e-05,
+      "loss": 0.0057,
+      "step": 15980
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.852027777777778e-05,
+      "loss": 0.0062,
+      "step": 15990
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.8519351851851854e-05,
+      "loss": 0.0076,
+      "step": 16000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.851842592592592e-05,
+      "loss": 0.0054,
+      "step": 16010
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.8517500000000004e-05,
+      "loss": 0.0074,
+      "step": 16020
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.851657407407408e-05,
+      "loss": 0.0081,
+      "step": 16030
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.8515648148148154e-05,
+      "loss": 0.0055,
+      "step": 16040
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.851472222222222e-05,
+      "loss": 0.0061,
+      "step": 16050
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.85137962962963e-05,
+      "loss": 0.0073,
+      "step": 16060
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.851287037037037e-05,
+      "loss": 0.0109,
+      "step": 16070
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.851194444444445e-05,
+      "loss": 0.0064,
+      "step": 16080
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.8511018518518516e-05,
+      "loss": 0.0067,
+      "step": 16090
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.851009259259259e-05,
+      "loss": 0.0067,
+      "step": 16100
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.8509166666666666e-05,
+      "loss": 0.0095,
+      "step": 16110
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.850824074074075e-05,
+      "loss": 0.0078,
+      "step": 16120
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.8507314814814816e-05,
+      "loss": 0.0065,
+      "step": 16130
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.850638888888889e-05,
+      "loss": 0.0076,
+      "step": 16140
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.8505462962962966e-05,
+      "loss": 0.0066,
+      "step": 16150
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.850453703703704e-05,
+      "loss": 0.0072,
+      "step": 16160
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.8503611111111117e-05,
+      "loss": 0.007,
+      "step": 16170
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.8502685185185185e-05,
+      "loss": 0.0089,
+      "step": 16180
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.850175925925926e-05,
+      "loss": 0.0116,
+      "step": 16190
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.8500833333333335e-05,
+      "loss": 0.0056,
+      "step": 16200
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 0.014792618341743946,
+      "eval_runtime": 153.0561,
+      "eval_samples_per_second": 26.069,
+      "eval_steps_per_second": 3.26,
+      "step": 16200
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.849990740740741e-05,
+      "loss": 0.0083,
+      "step": 16210
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.849898148148148e-05,
+      "loss": 0.0056,
+      "step": 16220
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.849805555555556e-05,
+      "loss": 0.0066,
+      "step": 16230
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.8497129629629635e-05,
+      "loss": 0.0065,
+      "step": 16240
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.849620370370371e-05,
+      "loss": 0.0077,
+      "step": 16250
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.849527777777778e-05,
+      "loss": 0.006,
+      "step": 16260
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.8494351851851854e-05,
+      "loss": 0.0047,
+      "step": 16270
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.849342592592593e-05,
+      "loss": 0.0045,
+      "step": 16280
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.8492500000000004e-05,
+      "loss": 0.0095,
+      "step": 16290
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.849157407407408e-05,
+      "loss": 0.0083,
+      "step": 16300
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.849064814814815e-05,
+      "loss": 0.0082,
+      "step": 16310
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.848972222222222e-05,
+      "loss": 0.0044,
+      "step": 16320
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.84887962962963e-05,
+      "loss": 0.0042,
+      "step": 16330
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.848787037037037e-05,
+      "loss": 0.0061,
+      "step": 16340
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.848694444444445e-05,
+      "loss": 0.0052,
+      "step": 16350
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.848601851851852e-05,
+      "loss": 0.007,
+      "step": 16360
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.84850925925926e-05,
+      "loss": 0.0067,
+      "step": 16370
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.848416666666667e-05,
+      "loss": 0.0092,
+      "step": 16380
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.848324074074074e-05,
+      "loss": 0.007,
+      "step": 16390
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.8482314814814816e-05,
+      "loss": 0.0096,
+      "step": 16400
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.848138888888889e-05,
+      "loss": 0.0074,
+      "step": 16410
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.8480462962962966e-05,
+      "loss": 0.0075,
+      "step": 16420
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.8479537037037034e-05,
+      "loss": 0.0034,
+      "step": 16430
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.847861111111111e-05,
+      "loss": 0.0075,
+      "step": 16440
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.847768518518519e-05,
+      "loss": 0.0087,
+      "step": 16450
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.8476759259259266e-05,
+      "loss": 0.0055,
+      "step": 16460
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.8475833333333334e-05,
+      "loss": 0.0094,
+      "step": 16470
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.847490740740741e-05,
+      "loss": 0.0038,
+      "step": 16480
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.8473981481481484e-05,
+      "loss": 0.0086,
+      "step": 16490
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.847305555555556e-05,
+      "loss": 0.0066,
+      "step": 16500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.8472129629629635e-05,
+      "loss": 0.0094,
+      "step": 16510
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.84712037037037e-05,
+      "loss": 0.0083,
+      "step": 16520
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.847027777777778e-05,
+      "loss": 0.0086,
+      "step": 16530
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.846935185185185e-05,
+      "loss": 0.0079,
+      "step": 16540
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.846842592592593e-05,
+      "loss": 0.006,
+      "step": 16550
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.84675e-05,
+      "loss": 0.0049,
+      "step": 16560
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.846657407407408e-05,
+      "loss": 0.0061,
+      "step": 16570
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.846564814814815e-05,
+      "loss": 0.0054,
+      "step": 16580
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.846472222222223e-05,
+      "loss": 0.0063,
+      "step": 16590
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.8463796296296297e-05,
+      "loss": 0.0048,
+      "step": 16600
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.846287037037037e-05,
+      "loss": 0.006,
+      "step": 16610
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.846194444444445e-05,
+      "loss": 0.0043,
+      "step": 16620
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.846101851851852e-05,
+      "loss": 0.0107,
+      "step": 16630
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.846009259259259e-05,
+      "loss": 0.0041,
+      "step": 16640
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.8459166666666665e-05,
+      "loss": 0.0042,
+      "step": 16650
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.845824074074075e-05,
+      "loss": 0.0079,
+      "step": 16660
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.845731481481482e-05,
+      "loss": 0.003,
+      "step": 16670
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.845638888888889e-05,
+      "loss": 0.0064,
+      "step": 16680
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.8455462962962965e-05,
+      "loss": 0.0058,
+      "step": 16690
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.845453703703704e-05,
+      "loss": 0.0059,
+      "step": 16700
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.8453611111111115e-05,
+      "loss": 0.0054,
+      "step": 16710
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.845268518518519e-05,
+      "loss": 0.0037,
+      "step": 16720
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.845175925925926e-05,
+      "loss": 0.007,
+      "step": 16730
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.8450833333333334e-05,
+      "loss": 0.009,
+      "step": 16740
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.844990740740741e-05,
+      "loss": 0.0052,
+      "step": 16750
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.8448981481481484e-05,
+      "loss": 0.0093,
+      "step": 16760
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.844805555555556e-05,
+      "loss": 0.0058,
+      "step": 16770
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.8447129629629634e-05,
+      "loss": 0.0075,
+      "step": 16780
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.844620370370371e-05,
+      "loss": 0.0076,
+      "step": 16790
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.8445277777777784e-05,
+      "loss": 0.0046,
+      "step": 16800
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.844435185185185e-05,
+      "loss": 0.0101,
+      "step": 16810
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.844342592592593e-05,
+      "loss": 0.0053,
+      "step": 16820
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.84425e-05,
+      "loss": 0.0061,
+      "step": 16830
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.844157407407408e-05,
+      "loss": 0.0044,
+      "step": 16840
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.8440648148148146e-05,
+      "loss": 0.0087,
+      "step": 16850
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.843972222222222e-05,
+      "loss": 0.0075,
+      "step": 16860
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.8438796296296296e-05,
+      "loss": 0.0072,
+      "step": 16870
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.843787037037038e-05,
+      "loss": 0.0065,
+      "step": 16880
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.8436944444444446e-05,
+      "loss": 0.0055,
+      "step": 16890
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.843601851851852e-05,
+      "loss": 0.0073,
+      "step": 16900
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.8435092592592596e-05,
+      "loss": 0.0075,
+      "step": 16910
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.843416666666667e-05,
+      "loss": 0.0059,
+      "step": 16920
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.8433240740740746e-05,
+      "loss": 0.0062,
+      "step": 16930
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.8432314814814815e-05,
+      "loss": 0.0057,
+      "step": 16940
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.843138888888889e-05,
+      "loss": 0.0074,
+      "step": 16950
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.8430462962962965e-05,
+      "loss": 0.0074,
+      "step": 16960
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.842953703703704e-05,
+      "loss": 0.0057,
+      "step": 16970
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.842861111111111e-05,
+      "loss": 0.0114,
+      "step": 16980
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.842768518518519e-05,
+      "loss": 0.0067,
+      "step": 16990
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.8426759259259265e-05,
+      "loss": 0.0082,
+      "step": 17000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.842583333333334e-05,
+      "loss": 0.0077,
+      "step": 17010
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.842490740740741e-05,
+      "loss": 0.0067,
+      "step": 17020
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.842398148148148e-05,
+      "loss": 0.0102,
+      "step": 17030
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.842305555555556e-05,
+      "loss": 0.0094,
+      "step": 17040
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.842212962962963e-05,
+      "loss": 0.0072,
+      "step": 17050
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.84212037037037e-05,
+      "loss": 0.0048,
+      "step": 17060
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.842027777777778e-05,
+      "loss": 0.0093,
+      "step": 17070
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.841935185185185e-05,
+      "loss": 0.0042,
+      "step": 17080
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.8418425925925934e-05,
+      "loss": 0.0097,
+      "step": 17090
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.84175e-05,
+      "loss": 0.0059,
+      "step": 17100
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.841657407407408e-05,
+      "loss": 0.0068,
+      "step": 17110
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.841564814814815e-05,
+      "loss": 0.008,
+      "step": 17120
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.841472222222223e-05,
+      "loss": 0.0061,
+      "step": 17130
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.84137962962963e-05,
+      "loss": 0.0064,
+      "step": 17140
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.841287037037037e-05,
+      "loss": 0.0037,
+      "step": 17150
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.8411944444444445e-05,
+      "loss": 0.0048,
+      "step": 17160
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.841101851851852e-05,
+      "loss": 0.0107,
+      "step": 17170
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.8410092592592596e-05,
+      "loss": 0.01,
+      "step": 17180
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.8409166666666664e-05,
+      "loss": 0.005,
+      "step": 17190
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.8408240740740746e-05,
+      "loss": 0.0098,
+      "step": 17200
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.840731481481482e-05,
+      "loss": 0.0035,
+      "step": 17210
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.8406388888888896e-05,
+      "loss": 0.0086,
+      "step": 17220
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.8405462962962964e-05,
+      "loss": 0.0057,
+      "step": 17230
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.840453703703704e-05,
+      "loss": 0.0024,
+      "step": 17240
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.8403611111111114e-05,
+      "loss": 0.0066,
+      "step": 17250
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.840268518518519e-05,
+      "loss": 0.0072,
+      "step": 17260
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.840175925925926e-05,
+      "loss": 0.0082,
+      "step": 17270
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.840083333333333e-05,
+      "loss": 0.0064,
+      "step": 17280
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.839990740740741e-05,
+      "loss": 0.0092,
+      "step": 17290
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.839898148148148e-05,
+      "loss": 0.0048,
+      "step": 17300
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.839805555555556e-05,
+      "loss": 0.0059,
+      "step": 17310
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.839712962962963e-05,
+      "loss": 0.005,
+      "step": 17320
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.839620370370371e-05,
+      "loss": 0.0093,
+      "step": 17330
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.839527777777778e-05,
+      "loss": 0.0049,
+      "step": 17340
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.839435185185186e-05,
+      "loss": 0.0044,
+      "step": 17350
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.8393425925925926e-05,
+      "loss": 0.0092,
+      "step": 17360
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.83925e-05,
+      "loss": 0.0069,
+      "step": 17370
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.8391574074074076e-05,
+      "loss": 0.0069,
+      "step": 17380
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.839064814814815e-05,
+      "loss": 0.0135,
+      "step": 17390
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.838972222222222e-05,
+      "loss": 0.0065,
+      "step": 17400
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.8388796296296295e-05,
+      "loss": 0.0057,
+      "step": 17410
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.8387870370370377e-05,
+      "loss": 0.0076,
+      "step": 17420
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.838694444444445e-05,
+      "loss": 0.0041,
+      "step": 17430
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.838601851851852e-05,
+      "loss": 0.0062,
+      "step": 17440
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.8385092592592595e-05,
+      "loss": 0.0054,
+      "step": 17450
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.838416666666667e-05,
+      "loss": 0.0068,
+      "step": 17460
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.8383240740740745e-05,
+      "loss": 0.0069,
+      "step": 17470
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.838231481481481e-05,
+      "loss": 0.0053,
+      "step": 17480
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.838138888888889e-05,
+      "loss": 0.0073,
+      "step": 17490
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.8380462962962963e-05,
+      "loss": 0.0058,
+      "step": 17500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.837953703703704e-05,
+      "loss": 0.006,
+      "step": 17510
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.8378611111111114e-05,
+      "loss": 0.0084,
+      "step": 17520
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.837768518518519e-05,
+      "loss": 0.0073,
+      "step": 17530
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.8376759259259264e-05,
+      "loss": 0.007,
+      "step": 17540
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.837583333333334e-05,
+      "loss": 0.0083,
+      "step": 17550
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.8374907407407414e-05,
+      "loss": 0.0078,
+      "step": 17560
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.837398148148148e-05,
+      "loss": 0.0071,
+      "step": 17570
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.837305555555556e-05,
+      "loss": 0.0074,
+      "step": 17580
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.837212962962963e-05,
+      "loss": 0.0081,
+      "step": 17590
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.837120370370371e-05,
+      "loss": 0.0073,
+      "step": 17600
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.8370277777777775e-05,
+      "loss": 0.0061,
+      "step": 17610
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.836935185185185e-05,
+      "loss": 0.011,
+      "step": 17620
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.836842592592593e-05,
+      "loss": 0.0059,
+      "step": 17630
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.836750000000001e-05,
+      "loss": 0.0088,
+      "step": 17640
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.8366574074074076e-05,
+      "loss": 0.0065,
+      "step": 17650
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.836564814814815e-05,
+      "loss": 0.007,
+      "step": 17660
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.8364722222222226e-05,
+      "loss": 0.0063,
+      "step": 17670
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.83637962962963e-05,
+      "loss": 0.0056,
+      "step": 17680
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.836287037037037e-05,
+      "loss": 0.0071,
+      "step": 17690
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.8361944444444444e-05,
+      "loss": 0.0053,
+      "step": 17700
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.836101851851852e-05,
+      "loss": 0.0051,
+      "step": 17710
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.8360092592592594e-05,
+      "loss": 0.0059,
+      "step": 17720
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.835916666666667e-05,
+      "loss": 0.0076,
+      "step": 17730
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.8358240740740744e-05,
+      "loss": 0.0082,
+      "step": 17740
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.835731481481482e-05,
+      "loss": 0.0055,
+      "step": 17750
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.8356388888888895e-05,
+      "loss": 0.0041,
+      "step": 17760
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.835546296296297e-05,
+      "loss": 0.0056,
+      "step": 17770
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.835453703703704e-05,
+      "loss": 0.0051,
+      "step": 17780
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.835361111111111e-05,
+      "loss": 0.0049,
+      "step": 17790
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.835268518518519e-05,
+      "loss": 0.0064,
+      "step": 17800
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.835175925925926e-05,
+      "loss": 0.0051,
+      "step": 17810
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.835083333333333e-05,
+      "loss": 0.0073,
+      "step": 17820
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.8349907407407406e-05,
+      "loss": 0.0076,
+      "step": 17830
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.834898148148148e-05,
+      "loss": 0.0086,
+      "step": 17840
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.834805555555556e-05,
+      "loss": 0.0075,
+      "step": 17850
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.834712962962963e-05,
+      "loss": 0.0065,
+      "step": 17860
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.8346203703703707e-05,
+      "loss": 0.0064,
+      "step": 17870
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.834527777777778e-05,
+      "loss": 0.0062,
+      "step": 17880
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.834435185185186e-05,
+      "loss": 0.0062,
+      "step": 17890
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.8343425925925925e-05,
+      "loss": 0.0053,
+      "step": 17900
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.83425e-05,
+      "loss": 0.0047,
+      "step": 17910
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.8341574074074075e-05,
+      "loss": 0.0087,
+      "step": 17920
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.834064814814815e-05,
+      "loss": 0.0072,
+      "step": 17930
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.8339722222222225e-05,
+      "loss": 0.0059,
+      "step": 17940
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.8338796296296293e-05,
+      "loss": 0.0064,
+      "step": 17950
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.8337870370370375e-05,
+      "loss": 0.0059,
+      "step": 17960
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.833694444444445e-05,
+      "loss": 0.0074,
+      "step": 17970
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.8336018518518525e-05,
+      "loss": 0.0085,
+      "step": 17980
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.8335092592592594e-05,
+      "loss": 0.0061,
+      "step": 17990
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.833416666666667e-05,
+      "loss": 0.0056,
+      "step": 18000
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.8333240740740744e-05,
+      "loss": 0.0061,
+      "step": 18010
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.833231481481482e-05,
+      "loss": 0.0085,
+      "step": 18020
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.833138888888889e-05,
+      "loss": 0.0061,
+      "step": 18030
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.833046296296296e-05,
+      "loss": 0.0066,
+      "step": 18040
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.832953703703704e-05,
+      "loss": 0.0054,
+      "step": 18050
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.832861111111112e-05,
+      "loss": 0.0103,
+      "step": 18060
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.832768518518519e-05,
+      "loss": 0.0112,
+      "step": 18070
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.832675925925926e-05,
+      "loss": 0.0048,
+      "step": 18080
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.832583333333334e-05,
+      "loss": 0.0055,
+      "step": 18090
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.832490740740741e-05,
+      "loss": 0.0062,
+      "step": 18100
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.832398148148148e-05,
+      "loss": 0.0049,
+      "step": 18110
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.8323055555555556e-05,
+      "loss": 0.0077,
+      "step": 18120
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.832212962962963e-05,
+      "loss": 0.0092,
+      "step": 18130
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.8321203703703706e-05,
+      "loss": 0.004,
+      "step": 18140
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.832027777777778e-05,
+      "loss": 0.0049,
+      "step": 18150
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.831935185185185e-05,
+      "loss": 0.0057,
+      "step": 18160
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.831842592592593e-05,
+      "loss": 0.0067,
+      "step": 18170
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.8317500000000006e-05,
+      "loss": 0.0078,
+      "step": 18180
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.831657407407408e-05,
+      "loss": 0.005,
+      "step": 18190
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.831564814814815e-05,
+      "loss": 0.0072,
+      "step": 18200
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.8314722222222225e-05,
+      "loss": 0.0059,
+      "step": 18210
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.83137962962963e-05,
+      "loss": 0.0049,
+      "step": 18220
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.8312870370370375e-05,
+      "loss": 0.006,
+      "step": 18230
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.831194444444444e-05,
+      "loss": 0.004,
+      "step": 18240
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.831101851851852e-05,
+      "loss": 0.0077,
+      "step": 18250
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.831009259259259e-05,
+      "loss": 0.0057,
+      "step": 18260
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.830916666666667e-05,
+      "loss": 0.0024,
+      "step": 18270
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.830824074074074e-05,
+      "loss": 0.0071,
+      "step": 18280
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.830731481481482e-05,
+      "loss": 0.004,
+      "step": 18290
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.830638888888889e-05,
+      "loss": 0.0076,
+      "step": 18300
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.830546296296297e-05,
+      "loss": 0.0076,
+      "step": 18310
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.830453703703704e-05,
+      "loss": 0.0054,
+      "step": 18320
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.830361111111111e-05,
+      "loss": 0.0033,
+      "step": 18330
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.830268518518519e-05,
+      "loss": 0.0065,
+      "step": 18340
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.830175925925926e-05,
+      "loss": 0.0055,
+      "step": 18350
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.830083333333334e-05,
+      "loss": 0.0088,
+      "step": 18360
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.8299907407407405e-05,
+      "loss": 0.0069,
+      "step": 18370
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.829898148148148e-05,
+      "loss": 0.0063,
+      "step": 18380
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.829805555555556e-05,
+      "loss": 0.0086,
+      "step": 18390
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.829712962962964e-05,
+      "loss": 0.0049,
+      "step": 18400
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8296203703703705e-05,
+      "loss": 0.0095,
+      "step": 18410
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.829527777777778e-05,
+      "loss": 0.0054,
+      "step": 18420
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8294351851851856e-05,
+      "loss": 0.0057,
+      "step": 18430
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.829342592592593e-05,
+      "loss": 0.0058,
+      "step": 18440
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.82925e-05,
+      "loss": 0.0096,
+      "step": 18450
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.8291574074074074e-05,
+      "loss": 0.0058,
+      "step": 18460
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.829064814814815e-05,
+      "loss": 0.0061,
+      "step": 18470
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.8289722222222224e-05,
+      "loss": 0.0046,
+      "step": 18480
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.82887962962963e-05,
+      "loss": 0.0042,
+      "step": 18490
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.8287870370370374e-05,
+      "loss": 0.0068,
+      "step": 18500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.828694444444445e-05,
+      "loss": 0.005,
+      "step": 18510
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.8286018518518524e-05,
+      "loss": 0.0061,
+      "step": 18520
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.828509259259259e-05,
+      "loss": 0.0062,
+      "step": 18530
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.828416666666667e-05,
+      "loss": 0.007,
+      "step": 18540
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.828324074074074e-05,
+      "loss": 0.0097,
+      "step": 18550
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.828231481481482e-05,
+      "loss": 0.0089,
+      "step": 18560
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.828138888888889e-05,
+      "loss": 0.008,
+      "step": 18570
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.828046296296296e-05,
+      "loss": 0.0058,
+      "step": 18580
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.8279537037037036e-05,
+      "loss": 0.0067,
+      "step": 18590
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.827861111111112e-05,
+      "loss": 0.0105,
+      "step": 18600
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.827768518518519e-05,
+      "loss": 0.0052,
+      "step": 18610
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.827675925925926e-05,
+      "loss": 0.0054,
+      "step": 18620
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.8275833333333336e-05,
+      "loss": 0.0033,
+      "step": 18630
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.827490740740741e-05,
+      "loss": 0.011,
+      "step": 18640
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.8273981481481486e-05,
+      "loss": 0.012,
+      "step": 18650
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.8273055555555555e-05,
+      "loss": 0.0062,
+      "step": 18660
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.827212962962963e-05,
+      "loss": 0.0078,
+      "step": 18670
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.8271203703703705e-05,
+      "loss": 0.0071,
+      "step": 18680
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.827027777777778e-05,
+      "loss": 0.0034,
+      "step": 18690
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.8269351851851855e-05,
+      "loss": 0.0092,
+      "step": 18700
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.826842592592593e-05,
+      "loss": 0.0091,
+      "step": 18710
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.8267500000000005e-05,
+      "loss": 0.0075,
+      "step": 18720
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.826657407407408e-05,
+      "loss": 0.0086,
+      "step": 18730
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.826564814814815e-05,
+      "loss": 0.0048,
+      "step": 18740
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.8264722222222223e-05,
+      "loss": 0.0052,
+      "step": 18750
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.82637962962963e-05,
+      "loss": 0.0089,
+      "step": 18760
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.8262870370370374e-05,
+      "loss": 0.0048,
+      "step": 18770
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.826194444444445e-05,
+      "loss": 0.0065,
+      "step": 18780
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.826101851851852e-05,
+      "loss": 0.0069,
+      "step": 18790
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.826009259259259e-05,
+      "loss": 0.0057,
+      "step": 18800
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.825916666666667e-05,
+      "loss": 0.0058,
+      "step": 18810
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.825824074074075e-05,
+      "loss": 0.0073,
+      "step": 18820
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.825731481481482e-05,
+      "loss": 0.0027,
+      "step": 18830
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.825638888888889e-05,
+      "loss": 0.0067,
+      "step": 18840
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.825546296296297e-05,
+      "loss": 0.0073,
+      "step": 18850
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.825453703703704e-05,
+      "loss": 0.0069,
+      "step": 18860
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.825361111111111e-05,
+      "loss": 0.0113,
+      "step": 18870
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.8252685185185186e-05,
+      "loss": 0.0055,
+      "step": 18880
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.825175925925926e-05,
+      "loss": 0.0082,
+      "step": 18890
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.8250833333333336e-05,
+      "loss": 0.0086,
+      "step": 18900
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.824990740740741e-05,
+      "loss": 0.0081,
+      "step": 18910
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.824898148148148e-05,
+      "loss": 0.0074,
+      "step": 18920
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.824805555555556e-05,
+      "loss": 0.0087,
+      "step": 18930
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.8247129629629636e-05,
+      "loss": 0.0079,
+      "step": 18940
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.8246203703703704e-05,
+      "loss": 0.0071,
+      "step": 18950
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.824527777777778e-05,
+      "loss": 0.0064,
+      "step": 18960
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.8244351851851854e-05,
+      "loss": 0.0046,
+      "step": 18970
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.824342592592593e-05,
+      "loss": 0.005,
+      "step": 18980
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.8242500000000004e-05,
+      "loss": 0.0069,
+      "step": 18990
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.824157407407407e-05,
+      "loss": 0.0078,
+      "step": 19000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.824064814814815e-05,
+      "loss": 0.0097,
+      "step": 19010
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.823972222222222e-05,
+      "loss": 0.0092,
+      "step": 19020
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.8238796296296305e-05,
+      "loss": 0.0071,
+      "step": 19030
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.823787037037037e-05,
+      "loss": 0.0088,
+      "step": 19040
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.823694444444445e-05,
+      "loss": 0.009,
+      "step": 19050
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.823601851851852e-05,
+      "loss": 0.0061,
+      "step": 19060
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.82350925925926e-05,
+      "loss": 0.0078,
+      "step": 19070
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.8234166666666666e-05,
+      "loss": 0.0117,
+      "step": 19080
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.823324074074074e-05,
+      "loss": 0.0065,
+      "step": 19090
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.8232314814814816e-05,
+      "loss": 0.008,
+      "step": 19100
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.823138888888889e-05,
+      "loss": 0.0066,
+      "step": 19110
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.8230462962962967e-05,
+      "loss": 0.0089,
+      "step": 19120
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.8229537037037035e-05,
+      "loss": 0.006,
+      "step": 19130
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.822861111111112e-05,
+      "loss": 0.0063,
+      "step": 19140
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.822768518518519e-05,
+      "loss": 0.0059,
+      "step": 19150
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.822675925925926e-05,
+      "loss": 0.0047,
+      "step": 19160
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.8225833333333335e-05,
+      "loss": 0.007,
+      "step": 19170
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.822490740740741e-05,
+      "loss": 0.0099,
+      "step": 19180
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.8223981481481485e-05,
+      "loss": 0.0057,
+      "step": 19190
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.822314814814815e-05,
+      "loss": 0.0089,
+      "step": 19200
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.8222222222222225e-05,
+      "loss": 0.0074,
+      "step": 19210
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.82212962962963e-05,
+      "loss": 0.005,
+      "step": 19220
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.822037037037037e-05,
+      "loss": 0.0076,
+      "step": 19230
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.8219444444444443e-05,
+      "loss": 0.0095,
+      "step": 19240
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.821851851851852e-05,
+      "loss": 0.0084,
+      "step": 19250
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.82175925925926e-05,
+      "loss": 0.0064,
+      "step": 19260
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.821666666666667e-05,
+      "loss": 0.0075,
+      "step": 19270
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.8215740740740744e-05,
+      "loss": 0.0063,
+      "step": 19280
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.821481481481482e-05,
+      "loss": 0.0058,
+      "step": 19290
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.8213888888888894e-05,
+      "loss": 0.0073,
+      "step": 19300
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.821296296296296e-05,
+      "loss": 0.0073,
+      "step": 19310
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.821203703703704e-05,
+      "loss": 0.006,
+      "step": 19320
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.821111111111111e-05,
+      "loss": 0.0048,
+      "step": 19330
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.821018518518519e-05,
+      "loss": 0.0059,
+      "step": 19340
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.820925925925926e-05,
+      "loss": 0.0034,
+      "step": 19350
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.820833333333333e-05,
+      "loss": 0.0048,
+      "step": 19360
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.820740740740741e-05,
+      "loss": 0.005,
+      "step": 19370
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.820648148148149e-05,
+      "loss": 0.0086,
+      "step": 19380
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.820555555555556e-05,
+      "loss": 0.0066,
+      "step": 19390
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.820462962962963e-05,
+      "loss": 0.0057,
+      "step": 19400
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.8203703703703706e-05,
+      "loss": 0.0082,
+      "step": 19410
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.820277777777778e-05,
+      "loss": 0.0038,
+      "step": 19420
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.8201851851851856e-05,
+      "loss": 0.0072,
+      "step": 19430
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.8200925925925924e-05,
+      "loss": 0.0074,
+      "step": 19440
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.82e-05,
+      "loss": 0.0064,
+      "step": 19450
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.8199074074074074e-05,
+      "loss": 0.0091,
+      "step": 19460
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.819814814814815e-05,
+      "loss": 0.0057,
+      "step": 19470
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.8197222222222224e-05,
+      "loss": 0.0079,
+      "step": 19480
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.81962962962963e-05,
+      "loss": 0.0078,
+      "step": 19490
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.8195370370370375e-05,
+      "loss": 0.0093,
+      "step": 19500
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.819444444444445e-05,
+      "loss": 0.0053,
+      "step": 19510
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.819351851851852e-05,
+      "loss": 0.0075,
+      "step": 19520
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.819259259259259e-05,
+      "loss": 0.0126,
+      "step": 19530
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.819166666666667e-05,
+      "loss": 0.0086,
+      "step": 19540
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.819074074074074e-05,
+      "loss": 0.0067,
+      "step": 19550
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.818981481481482e-05,
+      "loss": 0.0065,
+      "step": 19560
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.8188888888888886e-05,
+      "loss": 0.0045,
+      "step": 19570
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.818796296296296e-05,
+      "loss": 0.0033,
+      "step": 19580
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.818703703703704e-05,
+      "loss": 0.0089,
+      "step": 19590
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.818611111111112e-05,
+      "loss": 0.0075,
+      "step": 19600
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.818518518518519e-05,
+      "loss": 0.0041,
+      "step": 19610
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.818425925925926e-05,
+      "loss": 0.0067,
+      "step": 19620
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.818333333333334e-05,
+      "loss": 0.0062,
+      "step": 19630
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.818240740740741e-05,
+      "loss": 0.0053,
+      "step": 19640
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.818148148148148e-05,
+      "loss": 0.0099,
+      "step": 19650
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.8180555555555555e-05,
+      "loss": 0.0065,
+      "step": 19660
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.817962962962963e-05,
+      "loss": 0.0079,
+      "step": 19670
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.8178703703703705e-05,
+      "loss": 0.008,
+      "step": 19680
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.817777777777778e-05,
+      "loss": 0.0052,
+      "step": 19690
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.8176851851851855e-05,
+      "loss": 0.0045,
+      "step": 19700
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.817592592592593e-05,
+      "loss": 0.0062,
+      "step": 19710
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.8175000000000005e-05,
+      "loss": 0.0111,
+      "step": 19720
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.8174074074074074e-05,
+      "loss": 0.0048,
+      "step": 19730
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.817314814814815e-05,
+      "loss": 0.0043,
+      "step": 19740
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.8172222222222224e-05,
+      "loss": 0.0068,
+      "step": 19750
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.81712962962963e-05,
+      "loss": 0.0082,
+      "step": 19760
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.8170370370370374e-05,
+      "loss": 0.0074,
+      "step": 19770
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.816944444444444e-05,
+      "loss": 0.0116,
+      "step": 19780
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.816851851851852e-05,
+      "loss": 0.0051,
+      "step": 19790
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.81675925925926e-05,
+      "loss": 0.0046,
+      "step": 19800
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.8166666666666674e-05,
+      "loss": 0.0057,
+      "step": 19810
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.816574074074074e-05,
+      "loss": 0.0094,
+      "step": 19820
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.816481481481482e-05,
+      "loss": 0.0057,
+      "step": 19830
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.816388888888889e-05,
+      "loss": 0.0045,
+      "step": 19840
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.816296296296297e-05,
+      "loss": 0.0042,
+      "step": 19850
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.8162037037037036e-05,
+      "loss": 0.0074,
+      "step": 19860
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.816111111111111e-05,
+      "loss": 0.0074,
+      "step": 19870
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.8160185185185186e-05,
+      "loss": 0.008,
+      "step": 19880
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.815925925925926e-05,
+      "loss": 0.0049,
+      "step": 19890
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.8158333333333336e-05,
+      "loss": 0.0067,
+      "step": 19900
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.815740740740741e-05,
+      "loss": 0.0122,
+      "step": 19910
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.8156481481481486e-05,
+      "loss": 0.005,
+      "step": 19920
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.815555555555556e-05,
+      "loss": 0.0064,
+      "step": 19930
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.815462962962963e-05,
+      "loss": 0.0063,
+      "step": 19940
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.8153703703703705e-05,
+      "loss": 0.0077,
+      "step": 19950
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.815277777777778e-05,
+      "loss": 0.0057,
+      "step": 19960
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.8151851851851855e-05,
+      "loss": 0.0081,
+      "step": 19970
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.815092592592593e-05,
+      "loss": 0.0082,
+      "step": 19980
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.815e-05,
+      "loss": 0.0073,
+      "step": 19990
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.814907407407407e-05,
+      "loss": 0.0062,
+      "step": 20000
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.814814814814815e-05,
+      "loss": 0.0077,
+      "step": 20010
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.814722222222223e-05,
+      "loss": 0.0056,
+      "step": 20020
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.81462962962963e-05,
+      "loss": 0.0064,
+      "step": 20030
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.814537037037037e-05,
+      "loss": 0.0033,
+      "step": 20040
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.814444444444445e-05,
+      "loss": 0.0066,
+      "step": 20050
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.8143518518518523e-05,
+      "loss": 0.0059,
+      "step": 20060
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.814259259259259e-05,
+      "loss": 0.0103,
+      "step": 20070
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.814166666666667e-05,
+      "loss": 0.0088,
+      "step": 20080
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.814074074074074e-05,
+      "loss": 0.0073,
+      "step": 20090
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.813981481481482e-05,
+      "loss": 0.0064,
+      "step": 20100
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.813888888888889e-05,
+      "loss": 0.008,
+      "step": 20110
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.813796296296297e-05,
+      "loss": 0.0053,
+      "step": 20120
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.813703703703704e-05,
+      "loss": 0.0042,
+      "step": 20130
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.813611111111112e-05,
+      "loss": 0.0072,
+      "step": 20140
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.8135185185185185e-05,
+      "loss": 0.0051,
+      "step": 20150
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.813425925925926e-05,
+      "loss": 0.007,
+      "step": 20160
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.8133333333333336e-05,
+      "loss": 0.0067,
+      "step": 20170
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.813240740740741e-05,
+      "loss": 0.0124,
+      "step": 20180
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.8131481481481486e-05,
+      "loss": 0.0085,
+      "step": 20190
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.8130555555555554e-05,
+      "loss": 0.007,
+      "step": 20200
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.812962962962963e-05,
+      "loss": 0.0068,
+      "step": 20210
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.8128703703703704e-05,
+      "loss": 0.0069,
+      "step": 20220
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.8127777777777786e-05,
+      "loss": 0.0085,
+      "step": 20230
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.8126851851851854e-05,
+      "loss": 0.0059,
+      "step": 20240
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.812592592592593e-05,
+      "loss": 0.0069,
+      "step": 20250
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.8125000000000004e-05,
+      "loss": 0.0082,
+      "step": 20260
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.812407407407408e-05,
+      "loss": 0.0063,
+      "step": 20270
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.812314814814815e-05,
+      "loss": 0.0068,
+      "step": 20280
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.812222222222222e-05,
+      "loss": 0.0082,
+      "step": 20290
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.81212962962963e-05,
+      "loss": 0.0079,
+      "step": 20300
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.812037037037037e-05,
+      "loss": 0.004,
+      "step": 20310
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.811944444444445e-05,
+      "loss": 0.0087,
+      "step": 20320
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.8118518518518516e-05,
+      "loss": 0.0049,
+      "step": 20330
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.81175925925926e-05,
+      "loss": 0.0083,
+      "step": 20340
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.811666666666667e-05,
+      "loss": 0.0071,
+      "step": 20350
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.811574074074074e-05,
+      "loss": 0.0053,
+      "step": 20360
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.8114814814814816e-05,
+      "loss": 0.0068,
+      "step": 20370
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.811388888888889e-05,
+      "loss": 0.0046,
+      "step": 20380
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.8112962962962966e-05,
+      "loss": 0.0103,
+      "step": 20390
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.811203703703704e-05,
+      "loss": 0.0089,
+      "step": 20400
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.811111111111111e-05,
+      "loss": 0.0093,
+      "step": 20410
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.8110185185185185e-05,
+      "loss": 0.0069,
+      "step": 20420
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.810925925925926e-05,
+      "loss": 0.0066,
+      "step": 20430
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.8108333333333335e-05,
+      "loss": 0.0072,
+      "step": 20440
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.810740740740741e-05,
+      "loss": 0.0094,
+      "step": 20450
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.8106481481481485e-05,
+      "loss": 0.005,
+      "step": 20460
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.810555555555556e-05,
+      "loss": 0.0069,
+      "step": 20470
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.8104629629629635e-05,
+      "loss": 0.006,
+      "step": 20480
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.8103703703703703e-05,
+      "loss": 0.0048,
+      "step": 20490
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.810277777777778e-05,
+      "loss": 0.0064,
+      "step": 20500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.8101851851851854e-05,
+      "loss": 0.0049,
+      "step": 20510
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.810092592592593e-05,
+      "loss": 0.0104,
+      "step": 20520
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.8100000000000004e-05,
+      "loss": 0.0086,
+      "step": 20530
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.809907407407407e-05,
+      "loss": 0.0094,
+      "step": 20540
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.809814814814815e-05,
+      "loss": 0.0057,
+      "step": 20550
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.809722222222223e-05,
+      "loss": 0.0074,
+      "step": 20560
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.80962962962963e-05,
+      "loss": 0.0059,
+      "step": 20570
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.809537037037037e-05,
+      "loss": 0.0056,
+      "step": 20580
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.809444444444445e-05,
+      "loss": 0.0067,
+      "step": 20590
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.809351851851852e-05,
+      "loss": 0.0065,
+      "step": 20600
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.80925925925926e-05,
+      "loss": 0.0044,
+      "step": 20610
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.8091666666666666e-05,
+      "loss": 0.007,
+      "step": 20620
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.809074074074074e-05,
+      "loss": 0.0074,
+      "step": 20630
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.8089814814814816e-05,
+      "loss": 0.0051,
+      "step": 20640
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.808888888888889e-05,
+      "loss": 0.0062,
+      "step": 20650
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.8087962962962966e-05,
+      "loss": 0.0064,
+      "step": 20660
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.808703703703704e-05,
+      "loss": 0.007,
+      "step": 20670
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.8086111111111116e-05,
+      "loss": 0.0082,
+      "step": 20680
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.808518518518519e-05,
+      "loss": 0.0061,
+      "step": 20690
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.808425925925926e-05,
+      "loss": 0.0044,
+      "step": 20700
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.8083333333333334e-05,
+      "loss": 0.0056,
+      "step": 20710
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.808240740740741e-05,
+      "loss": 0.0062,
+      "step": 20720
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.8081481481481484e-05,
+      "loss": 0.0073,
+      "step": 20730
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.808055555555556e-05,
+      "loss": 0.0079,
+      "step": 20740
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.807962962962963e-05,
+      "loss": 0.0072,
+      "step": 20750
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.80787037037037e-05,
+      "loss": 0.0065,
+      "step": 20760
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.8077777777777785e-05,
+      "loss": 0.0051,
+      "step": 20770
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.807685185185185e-05,
+      "loss": 0.0067,
+      "step": 20780
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.807592592592593e-05,
+      "loss": 0.0106,
+      "step": 20790
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.8075e-05,
+      "loss": 0.0081,
+      "step": 20800
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.807407407407408e-05,
+      "loss": 0.0086,
+      "step": 20810
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.807314814814815e-05,
+      "loss": 0.0075,
+      "step": 20820
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.807222222222222e-05,
+      "loss": 0.0079,
+      "step": 20830
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.8071296296296296e-05,
+      "loss": 0.008,
+      "step": 20840
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.807037037037037e-05,
+      "loss": 0.0046,
+      "step": 20850
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.8069444444444447e-05,
+      "loss": 0.0098,
+      "step": 20860
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.806851851851852e-05,
+      "loss": 0.0061,
+      "step": 20870
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.80675925925926e-05,
+      "loss": 0.0065,
+      "step": 20880
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.806666666666667e-05,
+      "loss": 0.0092,
+      "step": 20890
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.806574074074075e-05,
+      "loss": 0.0044,
+      "step": 20900
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.8064814814814815e-05,
+      "loss": 0.008,
+      "step": 20910
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.806388888888889e-05,
+      "loss": 0.0079,
+      "step": 20920
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.8062962962962965e-05,
+      "loss": 0.0078,
+      "step": 20930
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.806203703703704e-05,
+      "loss": 0.0078,
+      "step": 20940
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.8061111111111115e-05,
+      "loss": 0.005,
+      "step": 20950
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.8060185185185184e-05,
+      "loss": 0.0062,
+      "step": 20960
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.805925925925926e-05,
+      "loss": 0.007,
+      "step": 20970
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.8058333333333334e-05,
+      "loss": 0.0081,
+      "step": 20980
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.805740740740741e-05,
+      "loss": 0.0068,
+      "step": 20990
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.8056481481481484e-05,
+      "loss": 0.008,
+      "step": 21000
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.805555555555556e-05,
+      "loss": 0.004,
+      "step": 21010
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.8054629629629634e-05,
+      "loss": 0.0062,
+      "step": 21020
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.805370370370371e-05,
+      "loss": 0.0045,
+      "step": 21030
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.805277777777778e-05,
+      "loss": 0.0078,
+      "step": 21040
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.805185185185185e-05,
+      "loss": 0.0074,
+      "step": 21050
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.805092592592593e-05,
+      "loss": 0.0065,
+      "step": 21060
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.805e-05,
+      "loss": 0.0051,
+      "step": 21070
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.804907407407408e-05,
+      "loss": 0.0044,
+      "step": 21080
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.8048148148148146e-05,
+      "loss": 0.0075,
+      "step": 21090
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.804722222222223e-05,
+      "loss": 0.0062,
+      "step": 21100
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.80462962962963e-05,
+      "loss": 0.007,
+      "step": 21110
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.804537037037037e-05,
+      "loss": 0.0066,
+      "step": 21120
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.8044444444444446e-05,
+      "loss": 0.0056,
+      "step": 21130
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.804351851851852e-05,
+      "loss": 0.0074,
+      "step": 21140
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.8042592592592596e-05,
+      "loss": 0.0061,
+      "step": 21150
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.804166666666667e-05,
+      "loss": 0.0053,
+      "step": 21160
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.804074074074074e-05,
+      "loss": 0.0102,
+      "step": 21170
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.8039814814814814e-05,
+      "loss": 0.007,
+      "step": 21180
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.803888888888889e-05,
+      "loss": 0.0069,
+      "step": 21190
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.8037962962962965e-05,
+      "loss": 0.0074,
+      "step": 21200
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.803703703703704e-05,
+      "loss": 0.0063,
+      "step": 21210
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.8036111111111115e-05,
+      "loss": 0.0075,
+      "step": 21220
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.803518518518519e-05,
+      "loss": 0.006,
+      "step": 21230
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.8034259259259265e-05,
+      "loss": 0.0047,
+      "step": 21240
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.803333333333333e-05,
+      "loss": 0.0056,
+      "step": 21250
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.803240740740741e-05,
+      "loss": 0.0045,
+      "step": 21260
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.803148148148148e-05,
+      "loss": 0.0059,
+      "step": 21270
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.803055555555556e-05,
+      "loss": 0.0068,
+      "step": 21280
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.802962962962963e-05,
+      "loss": 0.0076,
+      "step": 21290
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.80287037037037e-05,
+      "loss": 0.0073,
+      "step": 21300
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.8027777777777783e-05,
+      "loss": 0.0091,
+      "step": 21310
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.802694444444445e-05,
+      "loss": 0.0082,
+      "step": 21320
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.802601851851852e-05,
+      "loss": 0.0068,
+      "step": 21330
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.80250925925926e-05,
+      "loss": 0.0064,
+      "step": 21340
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.8024166666666673e-05,
+      "loss": 0.0084,
+      "step": 21350
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.802324074074074e-05,
+      "loss": 0.0058,
+      "step": 21360
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.802231481481482e-05,
+      "loss": 0.008,
+      "step": 21370
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.802138888888889e-05,
+      "loss": 0.0093,
+      "step": 21380
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.802046296296297e-05,
+      "loss": 0.0068,
+      "step": 21390
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.8019537037037035e-05,
+      "loss": 0.0049,
+      "step": 21400
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.801861111111111e-05,
+      "loss": 0.0054,
+      "step": 21410
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.8017685185185185e-05,
+      "loss": 0.0047,
+      "step": 21420
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.801675925925927e-05,
+      "loss": 0.0038,
+      "step": 21430
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.8015833333333335e-05,
+      "loss": 0.0064,
+      "step": 21440
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.801490740740741e-05,
+      "loss": 0.0054,
+      "step": 21450
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.8013981481481485e-05,
+      "loss": 0.0041,
+      "step": 21460
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.801305555555556e-05,
+      "loss": 0.0059,
+      "step": 21470
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.801212962962963e-05,
+      "loss": 0.0071,
+      "step": 21480
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.8011203703703704e-05,
+      "loss": 0.0061,
+      "step": 21490
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.801027777777778e-05,
+      "loss": 0.0048,
+      "step": 21500
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.8009351851851854e-05,
+      "loss": 0.0056,
+      "step": 21510
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.800842592592593e-05,
+      "loss": 0.0061,
+      "step": 21520
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.80075e-05,
+      "loss": 0.0063,
+      "step": 21530
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.800657407407408e-05,
+      "loss": 0.0079,
+      "step": 21540
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.8005648148148154e-05,
+      "loss": 0.0041,
+      "step": 21550
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.800472222222223e-05,
+      "loss": 0.0103,
+      "step": 21560
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.80037962962963e-05,
+      "loss": 0.0062,
+      "step": 21570
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.800287037037037e-05,
+      "loss": 0.004,
+      "step": 21580
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.800194444444445e-05,
+      "loss": 0.0104,
+      "step": 21590
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.800101851851852e-05,
+      "loss": 0.0071,
+      "step": 21600
+    },
+    {
+      "epoch": 4.0,
+      "eval_loss": 0.012618047185242176,
+      "eval_runtime": 153.9793,
+      "eval_samples_per_second": 25.913,
+      "eval_steps_per_second": 3.241,
+      "step": 21600
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.800009259259259e-05,
+      "loss": 0.0043,
+      "step": 21610
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.7999166666666666e-05,
+      "loss": 0.0052,
+      "step": 21620
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.799824074074074e-05,
+      "loss": 0.0063,
+      "step": 21630
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.7997314814814816e-05,
+      "loss": 0.0052,
+      "step": 21640
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.799638888888889e-05,
+      "loss": 0.0052,
+      "step": 21650
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.7995462962962966e-05,
+      "loss": 0.0046,
+      "step": 21660
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.799453703703704e-05,
+      "loss": 0.0095,
+      "step": 21670
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.7993611111111116e-05,
+      "loss": 0.0055,
+      "step": 21680
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.7992685185185185e-05,
+      "loss": 0.0065,
+      "step": 21690
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.799175925925926e-05,
+      "loss": 0.0067,
+      "step": 21700
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.7990833333333335e-05,
+      "loss": 0.0066,
+      "step": 21710
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.798990740740741e-05,
+      "loss": 0.0045,
+      "step": 21720
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.7988981481481485e-05,
+      "loss": 0.0053,
+      "step": 21730
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.798805555555555e-05,
+      "loss": 0.0056,
+      "step": 21740
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.7987129629629635e-05,
+      "loss": 0.0089,
+      "step": 21750
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.798620370370371e-05,
+      "loss": 0.0071,
+      "step": 21760
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.7985277777777785e-05,
+      "loss": 0.0068,
+      "step": 21770
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.7984351851851853e-05,
+      "loss": 0.0037,
+      "step": 21780
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.798342592592593e-05,
+      "loss": 0.0044,
+      "step": 21790
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.7982500000000004e-05,
+      "loss": 0.0079,
+      "step": 21800
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.798157407407408e-05,
+      "loss": 0.0092,
+      "step": 21810
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.798064814814815e-05,
+      "loss": 0.0065,
+      "step": 21820
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.797972222222222e-05,
+      "loss": 0.0068,
+      "step": 21830
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.79787962962963e-05,
+      "loss": 0.0058,
+      "step": 21840
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.797787037037037e-05,
+      "loss": 0.0035,
+      "step": 21850
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.797694444444445e-05,
+      "loss": 0.007,
+      "step": 21860
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.797601851851852e-05,
+      "loss": 0.0048,
+      "step": 21870
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.79750925925926e-05,
+      "loss": 0.0053,
+      "step": 21880
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.797416666666667e-05,
+      "loss": 0.0094,
+      "step": 21890
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.797324074074074e-05,
+      "loss": 0.0057,
+      "step": 21900
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.7972314814814816e-05,
+      "loss": 0.0074,
+      "step": 21910
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.797138888888889e-05,
+      "loss": 0.0096,
+      "step": 21920
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.7970462962962966e-05,
+      "loss": 0.0055,
+      "step": 21930
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.796953703703704e-05,
+      "loss": 0.0052,
+      "step": 21940
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.796861111111111e-05,
+      "loss": 0.0044,
+      "step": 21950
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.7967685185185184e-05,
+      "loss": 0.0047,
+      "step": 21960
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.7966759259259266e-05,
+      "loss": 0.0093,
+      "step": 21970
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.796583333333334e-05,
+      "loss": 0.0063,
+      "step": 21980
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.796490740740741e-05,
+      "loss": 0.0065,
+      "step": 21990
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.7963981481481484e-05,
+      "loss": 0.0069,
+      "step": 22000
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.796305555555556e-05,
+      "loss": 0.0065,
+      "step": 22010
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.7962129629629634e-05,
+      "loss": 0.0057,
+      "step": 22020
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.79612037037037e-05,
+      "loss": 0.0063,
+      "step": 22030
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.796027777777778e-05,
+      "loss": 0.0055,
+      "step": 22040
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.795935185185185e-05,
+      "loss": 0.0086,
+      "step": 22050
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.795842592592593e-05,
+      "loss": 0.0078,
+      "step": 22060
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.79575e-05,
+      "loss": 0.0083,
+      "step": 22070
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.795657407407408e-05,
+      "loss": 0.0067,
+      "step": 22080
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.795564814814815e-05,
+      "loss": 0.0061,
+      "step": 22090
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.795472222222223e-05,
+      "loss": 0.0088,
+      "step": 22100
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.7953796296296296e-05,
+      "loss": 0.0091,
+      "step": 22110
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.795287037037037e-05,
+      "loss": 0.009,
+      "step": 22120
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.7951944444444446e-05,
+      "loss": 0.0051,
+      "step": 22130
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.795101851851852e-05,
+      "loss": 0.0096,
+      "step": 22140
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.7950092592592597e-05,
+      "loss": 0.0067,
+      "step": 22150
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.7949166666666665e-05,
+      "loss": 0.0047,
+      "step": 22160
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.794824074074074e-05,
+      "loss": 0.0076,
+      "step": 22170
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.7947314814814815e-05,
+      "loss": 0.0069,
+      "step": 22180
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.79463888888889e-05,
+      "loss": 0.0065,
+      "step": 22190
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.7945462962962965e-05,
+      "loss": 0.0073,
+      "step": 22200
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.794453703703704e-05,
+      "loss": 0.0082,
+      "step": 22210
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.7943611111111115e-05,
+      "loss": 0.0087,
+      "step": 22220
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.794268518518519e-05,
+      "loss": 0.0041,
+      "step": 22230
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.794175925925926e-05,
+      "loss": 0.0058,
+      "step": 22240
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.7940833333333334e-05,
+      "loss": 0.0098,
+      "step": 22250
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.793990740740741e-05,
+      "loss": 0.007,
+      "step": 22260
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.7938981481481484e-05,
+      "loss": 0.0063,
+      "step": 22270
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.793805555555556e-05,
+      "loss": 0.0038,
+      "step": 22280
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.7937129629629634e-05,
+      "loss": 0.0062,
+      "step": 22290
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.793620370370371e-05,
+      "loss": 0.0043,
+      "step": 22300
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.7935277777777784e-05,
+      "loss": 0.0074,
+      "step": 22310
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.793435185185185e-05,
+      "loss": 0.0084,
+      "step": 22320
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.793342592592593e-05,
+      "loss": 0.0043,
+      "step": 22330
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.79325e-05,
+      "loss": 0.0038,
+      "step": 22340
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.793157407407408e-05,
+      "loss": 0.006,
+      "step": 22350
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.793064814814815e-05,
+      "loss": 0.0086,
+      "step": 22360
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.792972222222222e-05,
+      "loss": 0.0032,
+      "step": 22370
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.7928796296296296e-05,
+      "loss": 0.0042,
+      "step": 22380
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.792787037037037e-05,
+      "loss": 0.0047,
+      "step": 22390
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.792694444444445e-05,
+      "loss": 0.0051,
+      "step": 22400
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.792601851851852e-05,
+      "loss": 0.0054,
+      "step": 22410
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.7925092592592596e-05,
+      "loss": 0.0056,
+      "step": 22420
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.792416666666667e-05,
+      "loss": 0.0073,
+      "step": 22430
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.7923240740740746e-05,
+      "loss": 0.0049,
+      "step": 22440
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.7922314814814814e-05,
+      "loss": 0.0076,
+      "step": 22450
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.792138888888889e-05,
+      "loss": 0.0061,
+      "step": 22460
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.7920462962962964e-05,
+      "loss": 0.0061,
+      "step": 22470
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.791953703703704e-05,
+      "loss": 0.0079,
+      "step": 22480
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.7918611111111115e-05,
+      "loss": 0.0069,
+      "step": 22490
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.791768518518518e-05,
+      "loss": 0.0063,
+      "step": 22500
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.7916759259259265e-05,
+      "loss": 0.005,
+      "step": 22510
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.791583333333334e-05,
+      "loss": 0.0075,
+      "step": 22520
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.791490740740741e-05,
+      "loss": 0.0101,
+      "step": 22530
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.791398148148148e-05,
+      "loss": 0.0049,
+      "step": 22540
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.791305555555556e-05,
+      "loss": 0.0086,
+      "step": 22550
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.791212962962963e-05,
+      "loss": 0.0058,
+      "step": 22560
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.791120370370371e-05,
+      "loss": 0.0056,
+      "step": 22570
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.7910277777777777e-05,
+      "loss": 0.0083,
+      "step": 22580
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.790935185185185e-05,
+      "loss": 0.0055,
+      "step": 22590
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.790842592592593e-05,
+      "loss": 0.0075,
+      "step": 22600
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.79075e-05,
+      "loss": 0.0035,
+      "step": 22610
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.790657407407408e-05,
+      "loss": 0.0074,
+      "step": 22620
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.790564814814815e-05,
+      "loss": 0.0089,
+      "step": 22630
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.790472222222223e-05,
+      "loss": 0.0057,
+      "step": 22640
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.79037962962963e-05,
+      "loss": 0.0063,
+      "step": 22650
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.790287037037037e-05,
+      "loss": 0.0052,
+      "step": 22660
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.7901944444444445e-05,
+      "loss": 0.0049,
+      "step": 22670
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.790101851851852e-05,
+      "loss": 0.0069,
+      "step": 22680
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.7900092592592595e-05,
+      "loss": 0.0069,
+      "step": 22690
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.789916666666667e-05,
+      "loss": 0.0044,
+      "step": 22700
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.789824074074074e-05,
+      "loss": 0.0065,
+      "step": 22710
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.789731481481482e-05,
+      "loss": 0.0076,
+      "step": 22720
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.7896388888888896e-05,
+      "loss": 0.0092,
+      "step": 22730
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.7895462962962964e-05,
+      "loss": 0.0029,
+      "step": 22740
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.789453703703704e-05,
+      "loss": 0.0097,
+      "step": 22750
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.7893611111111114e-05,
+      "loss": 0.0073,
+      "step": 22760
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.789268518518519e-05,
+      "loss": 0.0079,
+      "step": 22770
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.7891759259259264e-05,
+      "loss": 0.006,
+      "step": 22780
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.789083333333333e-05,
+      "loss": 0.0082,
+      "step": 22790
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.788990740740741e-05,
+      "loss": 0.0067,
+      "step": 22800
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.788898148148148e-05,
+      "loss": 0.0034,
+      "step": 22810
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.788805555555556e-05,
+      "loss": 0.0043,
+      "step": 22820
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.788712962962963e-05,
+      "loss": 0.0044,
+      "step": 22830
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.788620370370371e-05,
+      "loss": 0.0068,
+      "step": 22840
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.788527777777778e-05,
+      "loss": 0.005,
+      "step": 22850
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.788435185185186e-05,
+      "loss": 0.0066,
+      "step": 22860
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.7883425925925926e-05,
+      "loss": 0.0038,
+      "step": 22870
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.78825e-05,
+      "loss": 0.0084,
+      "step": 22880
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.7881574074074076e-05,
+      "loss": 0.0074,
+      "step": 22890
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.788064814814815e-05,
+      "loss": 0.0034,
+      "step": 22900
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.7879722222222226e-05,
+      "loss": 0.0032,
+      "step": 22910
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.7878796296296295e-05,
+      "loss": 0.0072,
+      "step": 22920
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.787787037037037e-05,
+      "loss": 0.01,
+      "step": 22930
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.787694444444445e-05,
+      "loss": 0.0056,
+      "step": 22940
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.787601851851852e-05,
+      "loss": 0.0056,
+      "step": 22950
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.7875092592592595e-05,
+      "loss": 0.0059,
+      "step": 22960
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.787416666666667e-05,
+      "loss": 0.0095,
+      "step": 22970
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.7873240740740745e-05,
+      "loss": 0.0068,
+      "step": 22980
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.787231481481482e-05,
+      "loss": 0.0102,
+      "step": 22990
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.787138888888889e-05,
+      "loss": 0.0051,
+      "step": 23000
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.787046296296296e-05,
+      "loss": 0.0088,
+      "step": 23010
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.786953703703704e-05,
+      "loss": 0.0046,
+      "step": 23020
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.786861111111111e-05,
+      "loss": 0.0035,
+      "step": 23030
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.786768518518518e-05,
+      "loss": 0.0065,
+      "step": 23040
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.7866759259259263e-05,
+      "loss": 0.005,
+      "step": 23050
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.786583333333334e-05,
+      "loss": 0.0062,
+      "step": 23060
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.7864907407407414e-05,
+      "loss": 0.0078,
+      "step": 23070
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.786398148148148e-05,
+      "loss": 0.0062,
+      "step": 23080
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.786305555555556e-05,
+      "loss": 0.0043,
+      "step": 23090
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.786212962962963e-05,
+      "loss": 0.0039,
+      "step": 23100
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.786120370370371e-05,
+      "loss": 0.0061,
+      "step": 23110
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.786027777777778e-05,
+      "loss": 0.0038,
+      "step": 23120
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.785935185185185e-05,
+      "loss": 0.0049,
+      "step": 23130
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.7858425925925925e-05,
+      "loss": 0.007,
+      "step": 23140
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.78575e-05,
+      "loss": 0.0062,
+      "step": 23150
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.7856574074074076e-05,
+      "loss": 0.0066,
+      "step": 23160
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.785564814814815e-05,
+      "loss": 0.0043,
+      "step": 23170
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.7854722222222226e-05,
+      "loss": 0.0059,
+      "step": 23180
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.78537962962963e-05,
+      "loss": 0.0039,
+      "step": 23190
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.7852870370370376e-05,
+      "loss": 0.0055,
+      "step": 23200
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.7851944444444444e-05,
+      "loss": 0.0047,
+      "step": 23210
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.785101851851852e-05,
+      "loss": 0.0092,
+      "step": 23220
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.7850092592592594e-05,
+      "loss": 0.0046,
+      "step": 23230
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.784916666666667e-05,
+      "loss": 0.0059,
+      "step": 23240
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.784824074074074e-05,
+      "loss": 0.0063,
+      "step": 23250
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.784731481481482e-05,
+      "loss": 0.0041,
+      "step": 23260
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.7846388888888894e-05,
+      "loss": 0.0077,
+      "step": 23270
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.784546296296297e-05,
+      "loss": 0.0041,
+      "step": 23280
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.784453703703704e-05,
+      "loss": 0.0071,
+      "step": 23290
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.784361111111111e-05,
+      "loss": 0.0061,
+      "step": 23300
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.784268518518519e-05,
+      "loss": 0.0047,
+      "step": 23310
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.784175925925926e-05,
+      "loss": 0.0039,
+      "step": 23320
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.784083333333334e-05,
+      "loss": 0.005,
+      "step": 23330
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.7839907407407406e-05,
+      "loss": 0.0063,
+      "step": 23340
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.783898148148148e-05,
+      "loss": 0.0043,
+      "step": 23350
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.7838055555555556e-05,
+      "loss": 0.0036,
+      "step": 23360
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.783712962962963e-05,
+      "loss": 0.0065,
+      "step": 23370
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.7836203703703706e-05,
+      "loss": 0.0058,
+      "step": 23380
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.783527777777778e-05,
+      "loss": 0.0061,
+      "step": 23390
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.7834351851851857e-05,
+      "loss": 0.004,
+      "step": 23400
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.783342592592593e-05,
+      "loss": 0.0037,
+      "step": 23410
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.78325e-05,
+      "loss": 0.0071,
+      "step": 23420
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.7831574074074075e-05,
+      "loss": 0.0059,
+      "step": 23430
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.783064814814815e-05,
+      "loss": 0.0073,
+      "step": 23440
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.7829722222222225e-05,
+      "loss": 0.0078,
+      "step": 23450
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.782879629629629e-05,
+      "loss": 0.006,
+      "step": 23460
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.782787037037037e-05,
+      "loss": 0.0082,
+      "step": 23470
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.782694444444445e-05,
+      "loss": 0.0072,
+      "step": 23480
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.7826018518518525e-05,
+      "loss": 0.0114,
+      "step": 23490
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.7825092592592594e-05,
+      "loss": 0.005,
+      "step": 23500
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.782416666666667e-05,
+      "loss": 0.0047,
+      "step": 23510
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.7823240740740744e-05,
+      "loss": 0.0033,
+      "step": 23520
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.782231481481482e-05,
+      "loss": 0.0032,
+      "step": 23530
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.7821388888888894e-05,
+      "loss": 0.0055,
+      "step": 23540
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.782046296296296e-05,
+      "loss": 0.0054,
+      "step": 23550
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.781953703703704e-05,
+      "loss": 0.0065,
+      "step": 23560
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.781861111111111e-05,
+      "loss": 0.005,
+      "step": 23570
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.781768518518519e-05,
+      "loss": 0.0075,
+      "step": 23580
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.781675925925926e-05,
+      "loss": 0.0086,
+      "step": 23590
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.781583333333334e-05,
+      "loss": 0.0079,
+      "step": 23600
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.781490740740741e-05,
+      "loss": 0.0047,
+      "step": 23610
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.781398148148149e-05,
+      "loss": 0.006,
+      "step": 23620
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.7813055555555556e-05,
+      "loss": 0.0046,
+      "step": 23630
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.781212962962963e-05,
+      "loss": 0.0055,
+      "step": 23640
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.7811203703703706e-05,
+      "loss": 0.0053,
+      "step": 23650
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.781027777777778e-05,
+      "loss": 0.0051,
+      "step": 23660
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.780935185185185e-05,
+      "loss": 0.0067,
+      "step": 23670
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.7808425925925924e-05,
+      "loss": 0.005,
+      "step": 23680
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.7807500000000006e-05,
+      "loss": 0.0043,
+      "step": 23690
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.780657407407408e-05,
+      "loss": 0.005,
+      "step": 23700
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.780564814814815e-05,
+      "loss": 0.0071,
+      "step": 23710
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.7804722222222224e-05,
+      "loss": 0.0059,
+      "step": 23720
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.78037962962963e-05,
+      "loss": 0.0044,
+      "step": 23730
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.7802870370370375e-05,
+      "loss": 0.0033,
+      "step": 23740
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.780194444444445e-05,
+      "loss": 0.0073,
+      "step": 23750
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.780101851851852e-05,
+      "loss": 0.0039,
+      "step": 23760
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.780009259259259e-05,
+      "loss": 0.0093,
+      "step": 23770
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.779916666666667e-05,
+      "loss": 0.0057,
+      "step": 23780
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.779824074074074e-05,
+      "loss": 0.0055,
+      "step": 23790
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.779731481481482e-05,
+      "loss": 0.0067,
+      "step": 23800
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.779638888888889e-05,
+      "loss": 0.0069,
+      "step": 23810
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.779546296296297e-05,
+      "loss": 0.0075,
+      "step": 23820
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.779453703703704e-05,
+      "loss": 0.005,
+      "step": 23830
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.779361111111111e-05,
+      "loss": 0.0073,
+      "step": 23840
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.7792685185185187e-05,
+      "loss": 0.0071,
+      "step": 23850
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.779175925925926e-05,
+      "loss": 0.0078,
+      "step": 23860
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.779083333333334e-05,
+      "loss": 0.0065,
+      "step": 23870
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.7789907407407405e-05,
+      "loss": 0.0051,
+      "step": 23880
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.778898148148148e-05,
+      "loss": 0.0083,
+      "step": 23890
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.7788055555555555e-05,
+      "loss": 0.0057,
+      "step": 23900
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.778712962962964e-05,
+      "loss": 0.0054,
+      "step": 23910
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.7786203703703705e-05,
+      "loss": 0.0061,
+      "step": 23920
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.778527777777778e-05,
+      "loss": 0.0037,
+      "step": 23930
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.7784351851851855e-05,
+      "loss": 0.0072,
+      "step": 23940
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.778342592592593e-05,
+      "loss": 0.0087,
+      "step": 23950
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.7782500000000005e-05,
+      "loss": 0.0047,
+      "step": 23960
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.7781574074074074e-05,
+      "loss": 0.0067,
+      "step": 23970
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.778064814814815e-05,
+      "loss": 0.0062,
+      "step": 23980
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.7779722222222224e-05,
+      "loss": 0.0066,
+      "step": 23990
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.77787962962963e-05,
+      "loss": 0.0057,
+      "step": 24000
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.777787037037037e-05,
+      "loss": 0.0049,
+      "step": 24010
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.777694444444445e-05,
+      "loss": 0.0051,
+      "step": 24020
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.7776018518518524e-05,
+      "loss": 0.0071,
+      "step": 24030
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.77750925925926e-05,
+      "loss": 0.0038,
+      "step": 24040
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.777416666666667e-05,
+      "loss": 0.0048,
+      "step": 24050
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.777324074074074e-05,
+      "loss": 0.0049,
+      "step": 24060
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.777231481481482e-05,
+      "loss": 0.0051,
+      "step": 24070
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.777138888888889e-05,
+      "loss": 0.0049,
+      "step": 24080
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.777046296296296e-05,
+      "loss": 0.0076,
+      "step": 24090
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.7769537037037036e-05,
+      "loss": 0.0067,
+      "step": 24100
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.776861111111111e-05,
+      "loss": 0.0082,
+      "step": 24110
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.7767685185185186e-05,
+      "loss": 0.0046,
+      "step": 24120
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.776675925925926e-05,
+      "loss": 0.005,
+      "step": 24130
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.7765833333333336e-05,
+      "loss": 0.0057,
+      "step": 24140
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.776490740740741e-05,
+      "loss": 0.0048,
+      "step": 24150
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.7763981481481486e-05,
+      "loss": 0.0083,
+      "step": 24160
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.776305555555556e-05,
+      "loss": 0.0067,
+      "step": 24170
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.776212962962963e-05,
+      "loss": 0.0078,
+      "step": 24180
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.7761203703703705e-05,
+      "loss": 0.0043,
+      "step": 24190
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.776027777777778e-05,
+      "loss": 0.0077,
+      "step": 24200
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.7759351851851855e-05,
+      "loss": 0.005,
+      "step": 24210
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.775842592592592e-05,
+      "loss": 0.0066,
+      "step": 24220
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.7757500000000005e-05,
+      "loss": 0.0049,
+      "step": 24230
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.775657407407408e-05,
+      "loss": 0.0061,
+      "step": 24240
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.7755648148148155e-05,
+      "loss": 0.0086,
+      "step": 24250
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.775472222222222e-05,
+      "loss": 0.007,
+      "step": 24260
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.77537962962963e-05,
+      "loss": 0.0046,
+      "step": 24270
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.775287037037037e-05,
+      "loss": 0.007,
+      "step": 24280
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.775194444444445e-05,
+      "loss": 0.0072,
+      "step": 24290
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.775101851851852e-05,
+      "loss": 0.0067,
+      "step": 24300
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.775009259259259e-05,
+      "loss": 0.0088,
+      "step": 24310
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.774916666666667e-05,
+      "loss": 0.0057,
+      "step": 24320
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.774824074074074e-05,
+      "loss": 0.0051,
+      "step": 24330
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.774731481481482e-05,
+      "loss": 0.0051,
+      "step": 24340
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.774638888888889e-05,
+      "loss": 0.0051,
+      "step": 24350
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.774546296296297e-05,
+      "loss": 0.0056,
+      "step": 24360
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.774453703703704e-05,
+      "loss": 0.0075,
+      "step": 24370
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.774361111111112e-05,
+      "loss": 0.0044,
+      "step": 24380
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7742685185185185e-05,
+      "loss": 0.0074,
+      "step": 24390
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.774175925925926e-05,
+      "loss": 0.0068,
+      "step": 24400
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7740833333333335e-05,
+      "loss": 0.0051,
+      "step": 24410
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.773990740740741e-05,
+      "loss": 0.0096,
+      "step": 24420
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.773898148148148e-05,
+      "loss": 0.0048,
+      "step": 24430
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.7738055555555554e-05,
+      "loss": 0.0038,
+      "step": 24440
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.7737129629629636e-05,
+      "loss": 0.0076,
+      "step": 24450
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.773620370370371e-05,
+      "loss": 0.004,
+      "step": 24460
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.773527777777778e-05,
+      "loss": 0.0044,
+      "step": 24470
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.7734351851851854e-05,
+      "loss": 0.0061,
+      "step": 24480
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.773342592592593e-05,
+      "loss": 0.0045,
+      "step": 24490
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.7732500000000004e-05,
+      "loss": 0.0052,
+      "step": 24500
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.773157407407407e-05,
+      "loss": 0.0077,
+      "step": 24510
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.773064814814815e-05,
+      "loss": 0.0042,
+      "step": 24520
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.772972222222222e-05,
+      "loss": 0.0074,
+      "step": 24530
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.77287962962963e-05,
+      "loss": 0.005,
+      "step": 24540
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.772787037037037e-05,
+      "loss": 0.0033,
+      "step": 24550
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.772694444444445e-05,
+      "loss": 0.0045,
+      "step": 24560
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.772611111111111e-05,
+      "loss": 0.0088,
+      "step": 24570
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.772518518518519e-05,
+      "loss": 0.0078,
+      "step": 24580
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.772425925925926e-05,
+      "loss": 0.0044,
+      "step": 24590
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.772333333333334e-05,
+      "loss": 0.0071,
+      "step": 24600
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.772240740740741e-05,
+      "loss": 0.0072,
+      "step": 24610
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.772148148148148e-05,
+      "loss": 0.007,
+      "step": 24620
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.7720555555555556e-05,
+      "loss": 0.004,
+      "step": 24630
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.771962962962963e-05,
+      "loss": 0.0039,
+      "step": 24640
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.7718703703703706e-05,
+      "loss": 0.0036,
+      "step": 24650
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.7717777777777775e-05,
+      "loss": 0.0074,
+      "step": 24660
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.771685185185185e-05,
+      "loss": 0.0044,
+      "step": 24670
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.771592592592593e-05,
+      "loss": 0.0076,
+      "step": 24680
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.7715000000000006e-05,
+      "loss": 0.006,
+      "step": 24690
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.7714074074074075e-05,
+      "loss": 0.0059,
+      "step": 24700
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.771314814814815e-05,
+      "loss": 0.0053,
+      "step": 24710
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.7712222222222225e-05,
+      "loss": 0.0064,
+      "step": 24720
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.77112962962963e-05,
+      "loss": 0.0047,
+      "step": 24730
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.7710370370370375e-05,
+      "loss": 0.0044,
+      "step": 24740
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.770944444444444e-05,
+      "loss": 0.0074,
+      "step": 24750
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.770851851851852e-05,
+      "loss": 0.0042,
+      "step": 24760
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.770759259259259e-05,
+      "loss": 0.0061,
+      "step": 24770
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.770666666666667e-05,
+      "loss": 0.0052,
+      "step": 24780
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.7705740740740743e-05,
+      "loss": 0.0062,
+      "step": 24790
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.770481481481482e-05,
+      "loss": 0.0049,
+      "step": 24800
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.7703888888888894e-05,
+      "loss": 0.0053,
+      "step": 24810
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.770296296296297e-05,
+      "loss": 0.0072,
+      "step": 24820
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.770203703703704e-05,
+      "loss": 0.0059,
+      "step": 24830
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.770111111111111e-05,
+      "loss": 0.005,
+      "step": 24840
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.770018518518519e-05,
+      "loss": 0.0079,
+      "step": 24850
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.769925925925926e-05,
+      "loss": 0.0066,
+      "step": 24860
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.769833333333333e-05,
+      "loss": 0.0069,
+      "step": 24870
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.7697407407407405e-05,
+      "loss": 0.0051,
+      "step": 24880
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.769648148148149e-05,
+      "loss": 0.0073,
+      "step": 24890
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.769555555555556e-05,
+      "loss": 0.0049,
+      "step": 24900
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.769462962962963e-05,
+      "loss": 0.0055,
+      "step": 24910
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.7693703703703706e-05,
+      "loss": 0.0059,
+      "step": 24920
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.769277777777778e-05,
+      "loss": 0.007,
+      "step": 24930
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.7691851851851856e-05,
+      "loss": 0.0063,
+      "step": 24940
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.769092592592593e-05,
+      "loss": 0.0052,
+      "step": 24950
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.769e-05,
+      "loss": 0.0061,
+      "step": 24960
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.7689074074074074e-05,
+      "loss": 0.0065,
+      "step": 24970
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.768814814814815e-05,
+      "loss": 0.009,
+      "step": 24980
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.7687222222222224e-05,
+      "loss": 0.0072,
+      "step": 24990
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.76862962962963e-05,
+      "loss": 0.0052,
+      "step": 25000
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.7685370370370374e-05,
+      "loss": 0.0042,
+      "step": 25010
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.768444444444445e-05,
+      "loss": 0.009,
+      "step": 25020
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.7683518518518524e-05,
+      "loss": 0.0045,
+      "step": 25030
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.768259259259259e-05,
+      "loss": 0.0044,
+      "step": 25040
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.768166666666667e-05,
+      "loss": 0.0051,
+      "step": 25050
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.768074074074074e-05,
+      "loss": 0.0046,
+      "step": 25060
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.767981481481482e-05,
+      "loss": 0.0104,
+      "step": 25070
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.767888888888889e-05,
+      "loss": 0.0083,
+      "step": 25080
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.767796296296296e-05,
+      "loss": 0.0065,
+      "step": 25090
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.7677037037037036e-05,
+      "loss": 0.0096,
+      "step": 25100
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.767611111111112e-05,
+      "loss": 0.0044,
+      "step": 25110
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.7675185185185186e-05,
+      "loss": 0.0057,
+      "step": 25120
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.767425925925926e-05,
+      "loss": 0.0075,
+      "step": 25130
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.7673333333333337e-05,
+      "loss": 0.0048,
+      "step": 25140
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.767240740740741e-05,
+      "loss": 0.0061,
+      "step": 25150
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.767148148148149e-05,
+      "loss": 0.0047,
+      "step": 25160
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.7670555555555555e-05,
+      "loss": 0.0071,
+      "step": 25170
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.766962962962963e-05,
+      "loss": 0.0062,
+      "step": 25180
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.7668703703703705e-05,
+      "loss": 0.0061,
+      "step": 25190
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.766777777777778e-05,
+      "loss": 0.007,
+      "step": 25200
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.766685185185185e-05,
+      "loss": 0.0049,
+      "step": 25210
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.766592592592593e-05,
+      "loss": 0.0063,
+      "step": 25220
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.7665000000000005e-05,
+      "loss": 0.0061,
+      "step": 25230
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.766407407407408e-05,
+      "loss": 0.0052,
+      "step": 25240
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.766314814814815e-05,
+      "loss": 0.007,
+      "step": 25250
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.7662222222222224e-05,
+      "loss": 0.0083,
+      "step": 25260
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.76612962962963e-05,
+      "loss": 0.0072,
+      "step": 25270
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.7660370370370374e-05,
+      "loss": 0.0057,
+      "step": 25280
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.765944444444445e-05,
+      "loss": 0.0046,
+      "step": 25290
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.765851851851852e-05,
+      "loss": 0.0052,
+      "step": 25300
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.765759259259259e-05,
+      "loss": 0.003,
+      "step": 25310
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.7656666666666674e-05,
+      "loss": 0.0045,
+      "step": 25320
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.765574074074074e-05,
+      "loss": 0.0069,
+      "step": 25330
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.765481481481482e-05,
+      "loss": 0.0047,
+      "step": 25340
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.765388888888889e-05,
+      "loss": 0.0055,
+      "step": 25350
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.765296296296297e-05,
+      "loss": 0.0053,
+      "step": 25360
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.765203703703704e-05,
+      "loss": 0.0063,
+      "step": 25370
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.765111111111111e-05,
+      "loss": 0.005,
+      "step": 25380
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.7650185185185186e-05,
+      "loss": 0.0036,
+      "step": 25390
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.764925925925926e-05,
+      "loss": 0.0039,
+      "step": 25400
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.7648333333333336e-05,
+      "loss": 0.0084,
+      "step": 25410
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.7647407407407404e-05,
+      "loss": 0.0037,
+      "step": 25420
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.7646481481481486e-05,
+      "loss": 0.0075,
+      "step": 25430
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.764555555555556e-05,
+      "loss": 0.0037,
+      "step": 25440
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.7644629629629636e-05,
+      "loss": 0.0054,
+      "step": 25450
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.7643703703703704e-05,
+      "loss": 0.0087,
+      "step": 25460
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.764277777777778e-05,
+      "loss": 0.0048,
+      "step": 25470
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.7641851851851855e-05,
+      "loss": 0.0042,
+      "step": 25480
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.764092592592593e-05,
+      "loss": 0.0082,
+      "step": 25490
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.7640000000000005e-05,
+      "loss": 0.0033,
+      "step": 25500
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.763907407407407e-05,
+      "loss": 0.0075,
+      "step": 25510
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.763814814814815e-05,
+      "loss": 0.0053,
+      "step": 25520
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.763722222222222e-05,
+      "loss": 0.0067,
+      "step": 25530
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.76362962962963e-05,
+      "loss": 0.007,
+      "step": 25540
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.763537037037037e-05,
+      "loss": 0.0053,
+      "step": 25550
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.763444444444445e-05,
+      "loss": 0.008,
+      "step": 25560
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.763351851851852e-05,
+      "loss": 0.0117,
+      "step": 25570
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.76325925925926e-05,
+      "loss": 0.006,
+      "step": 25580
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.7631666666666667e-05,
+      "loss": 0.0038,
+      "step": 25590
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.763074074074074e-05,
+      "loss": 0.0049,
+      "step": 25600
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.762981481481482e-05,
+      "loss": 0.0065,
+      "step": 25610
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.762888888888889e-05,
+      "loss": 0.006,
+      "step": 25620
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.762796296296296e-05,
+      "loss": 0.0058,
+      "step": 25630
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.7627037037037035e-05,
+      "loss": 0.0068,
+      "step": 25640
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.762611111111112e-05,
+      "loss": 0.0073,
+      "step": 25650
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.762518518518519e-05,
+      "loss": 0.0058,
+      "step": 25660
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.762425925925926e-05,
+      "loss": 0.0034,
+      "step": 25670
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.7623333333333335e-05,
+      "loss": 0.0068,
+      "step": 25680
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.762240740740741e-05,
+      "loss": 0.0074,
+      "step": 25690
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.7621481481481485e-05,
+      "loss": 0.0062,
+      "step": 25700
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.762055555555556e-05,
+      "loss": 0.004,
+      "step": 25710
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.761962962962963e-05,
+      "loss": 0.007,
+      "step": 25720
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.7618703703703704e-05,
+      "loss": 0.0063,
+      "step": 25730
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.761777777777778e-05,
+      "loss": 0.0061,
+      "step": 25740
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.7616851851851854e-05,
+      "loss": 0.0077,
+      "step": 25750
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.761592592592593e-05,
+      "loss": 0.0059,
+      "step": 25760
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.7615000000000004e-05,
+      "loss": 0.0104,
+      "step": 25770
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.761407407407408e-05,
+      "loss": 0.0062,
+      "step": 25780
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.7613148148148154e-05,
+      "loss": 0.0056,
+      "step": 25790
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.761222222222222e-05,
+      "loss": 0.0057,
+      "step": 25800
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.76112962962963e-05,
+      "loss": 0.0058,
+      "step": 25810
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.761037037037037e-05,
+      "loss": 0.0042,
+      "step": 25820
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.760944444444445e-05,
+      "loss": 0.0043,
+      "step": 25830
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.7608518518518516e-05,
+      "loss": 0.0086,
+      "step": 25840
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.760759259259259e-05,
+      "loss": 0.0072,
+      "step": 25850
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.760666666666667e-05,
+      "loss": 0.0051,
+      "step": 25860
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.760574074074075e-05,
+      "loss": 0.0059,
+      "step": 25870
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.7604814814814816e-05,
+      "loss": 0.0042,
+      "step": 25880
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.760388888888889e-05,
+      "loss": 0.0071,
+      "step": 25890
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.7602962962962966e-05,
+      "loss": 0.0058,
+      "step": 25900
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.760203703703704e-05,
+      "loss": 0.0065,
+      "step": 25910
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.7601111111111116e-05,
+      "loss": 0.0055,
+      "step": 25920
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.7600185185185185e-05,
+      "loss": 0.0055,
+      "step": 25930
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.759925925925926e-05,
+      "loss": 0.0056,
+      "step": 25940
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.7598333333333335e-05,
+      "loss": 0.0033,
+      "step": 25950
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.759740740740741e-05,
+      "loss": 0.004,
+      "step": 25960
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.7596481481481485e-05,
+      "loss": 0.0042,
+      "step": 25970
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.759555555555556e-05,
+      "loss": 0.0081,
+      "step": 25980
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.7594629629629635e-05,
+      "loss": 0.004,
+      "step": 25990
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.759370370370371e-05,
+      "loss": 0.0063,
+      "step": 26000
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.759277777777778e-05,
+      "loss": 0.0069,
+      "step": 26010
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.759185185185185e-05,
+      "loss": 0.0052,
+      "step": 26020
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.759092592592593e-05,
+      "loss": 0.0047,
+      "step": 26030
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.7590000000000003e-05,
+      "loss": 0.0072,
+      "step": 26040
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 4.758907407407407e-05,
+      "loss": 0.0069,
+      "step": 26050
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.758814814814815e-05,
+      "loss": 0.006,
+      "step": 26060
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.758722222222222e-05,
+      "loss": 0.0062,
+      "step": 26070
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.7586296296296304e-05,
+      "loss": 0.0051,
+      "step": 26080
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.758537037037037e-05,
+      "loss": 0.0056,
+      "step": 26090
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.758444444444445e-05,
+      "loss": 0.0054,
+      "step": 26100
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.758351851851852e-05,
+      "loss": 0.0047,
+      "step": 26110
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.75825925925926e-05,
+      "loss": 0.006,
+      "step": 26120
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.758166666666667e-05,
+      "loss": 0.0096,
+      "step": 26130
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.758074074074074e-05,
+      "loss": 0.0061,
+      "step": 26140
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.7579814814814816e-05,
+      "loss": 0.0057,
+      "step": 26150
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.757888888888889e-05,
+      "loss": 0.0088,
+      "step": 26160
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.7577962962962966e-05,
+      "loss": 0.0045,
+      "step": 26170
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.7577037037037034e-05,
+      "loss": 0.0069,
+      "step": 26180
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.7576111111111116e-05,
+      "loss": 0.0068,
+      "step": 26190
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.757518518518519e-05,
+      "loss": 0.0072,
+      "step": 26200
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.7574259259259266e-05,
+      "loss": 0.0049,
+      "step": 26210
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.7573333333333334e-05,
+      "loss": 0.0077,
+      "step": 26220
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.757240740740741e-05,
+      "loss": 0.0068,
+      "step": 26230
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.7571481481481484e-05,
+      "loss": 0.0048,
+      "step": 26240
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.757055555555556e-05,
+      "loss": 0.0034,
+      "step": 26250
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.756962962962963e-05,
+      "loss": 0.0085,
+      "step": 26260
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.75687037037037e-05,
+      "loss": 0.0047,
+      "step": 26270
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.756777777777778e-05,
+      "loss": 0.009,
+      "step": 26280
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.756685185185186e-05,
+      "loss": 0.0063,
+      "step": 26290
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.756592592592593e-05,
+      "loss": 0.0083,
+      "step": 26300
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.7565e-05,
+      "loss": 0.0066,
+      "step": 26310
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.756407407407408e-05,
+      "loss": 0.0063,
+      "step": 26320
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.756314814814815e-05,
+      "loss": 0.0081,
+      "step": 26330
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.756222222222223e-05,
+      "loss": 0.0074,
+      "step": 26340
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.7561296296296296e-05,
+      "loss": 0.0059,
+      "step": 26350
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.756037037037037e-05,
+      "loss": 0.0075,
+      "step": 26360
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.7559444444444446e-05,
+      "loss": 0.0053,
+      "step": 26370
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.755851851851852e-05,
+      "loss": 0.0055,
+      "step": 26380
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.755759259259259e-05,
+      "loss": 0.0075,
+      "step": 26390
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.755666666666667e-05,
+      "loss": 0.0039,
+      "step": 26400
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.755574074074075e-05,
+      "loss": 0.0043,
+      "step": 26410
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.755481481481482e-05,
+      "loss": 0.0051,
+      "step": 26420
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.755388888888889e-05,
+      "loss": 0.0062,
+      "step": 26430
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.7552962962962965e-05,
+      "loss": 0.0057,
+      "step": 26440
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.755203703703704e-05,
+      "loss": 0.0079,
+      "step": 26450
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.7551111111111115e-05,
+      "loss": 0.0069,
+      "step": 26460
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.7550185185185183e-05,
+      "loss": 0.0053,
+      "step": 26470
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.754925925925926e-05,
+      "loss": 0.0051,
+      "step": 26480
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.7548333333333334e-05,
+      "loss": 0.0059,
+      "step": 26490
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.754740740740741e-05,
+      "loss": 0.0048,
+      "step": 26500
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.7546481481481484e-05,
+      "loss": 0.0057,
+      "step": 26510
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.754555555555556e-05,
+      "loss": 0.0049,
+      "step": 26520
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.7544629629629634e-05,
+      "loss": 0.0053,
+      "step": 26530
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.754370370370371e-05,
+      "loss": 0.0086,
+      "step": 26540
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.7542777777777784e-05,
+      "loss": 0.0051,
+      "step": 26550
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.754185185185185e-05,
+      "loss": 0.005,
+      "step": 26560
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.754092592592593e-05,
+      "loss": 0.007,
+      "step": 26570
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.754e-05,
+      "loss": 0.0057,
+      "step": 26580
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.753907407407408e-05,
+      "loss": 0.0065,
+      "step": 26590
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.7538148148148146e-05,
+      "loss": 0.0077,
+      "step": 26600
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.753722222222222e-05,
+      "loss": 0.0063,
+      "step": 26610
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.75362962962963e-05,
+      "loss": 0.0069,
+      "step": 26620
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.753537037037038e-05,
+      "loss": 0.0065,
+      "step": 26630
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.7534444444444446e-05,
+      "loss": 0.0059,
+      "step": 26640
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.753351851851852e-05,
+      "loss": 0.0062,
+      "step": 26650
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.7532592592592596e-05,
+      "loss": 0.0082,
+      "step": 26660
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.753166666666667e-05,
+      "loss": 0.0062,
+      "step": 26670
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.753074074074074e-05,
+      "loss": 0.0068,
+      "step": 26680
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.7529814814814814e-05,
+      "loss": 0.006,
+      "step": 26690
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.752888888888889e-05,
+      "loss": 0.0057,
+      "step": 26700
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.7527962962962964e-05,
+      "loss": 0.0041,
+      "step": 26710
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.752703703703704e-05,
+      "loss": 0.0055,
+      "step": 26720
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.7526111111111115e-05,
+      "loss": 0.0042,
+      "step": 26730
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.752518518518519e-05,
+      "loss": 0.0072,
+      "step": 26740
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.7524259259259265e-05,
+      "loss": 0.0044,
+      "step": 26750
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.752333333333334e-05,
+      "loss": 0.0063,
+      "step": 26760
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.752240740740741e-05,
+      "loss": 0.0042,
+      "step": 26770
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.752148148148148e-05,
+      "loss": 0.0073,
+      "step": 26780
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.752055555555556e-05,
+      "loss": 0.0076,
+      "step": 26790
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.751962962962963e-05,
+      "loss": 0.0038,
+      "step": 26800
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.75187037037037e-05,
+      "loss": 0.005,
+      "step": 26810
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.7517777777777776e-05,
+      "loss": 0.0066,
+      "step": 26820
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.751685185185186e-05,
+      "loss": 0.0049,
+      "step": 26830
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.751592592592593e-05,
+      "loss": 0.006,
+      "step": 26840
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.7515e-05,
+      "loss": 0.0066,
+      "step": 26850
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.751407407407408e-05,
+      "loss": 0.0052,
+      "step": 26860
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.751314814814815e-05,
+      "loss": 0.0062,
+      "step": 26870
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.751222222222223e-05,
+      "loss": 0.0081,
+      "step": 26880
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.7511296296296295e-05,
+      "loss": 0.005,
+      "step": 26890
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.751037037037037e-05,
+      "loss": 0.0034,
+      "step": 26900
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.7509444444444445e-05,
+      "loss": 0.0075,
+      "step": 26910
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.750851851851852e-05,
+      "loss": 0.0072,
+      "step": 26920
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.7507592592592595e-05,
+      "loss": 0.0068,
+      "step": 26930
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.750666666666667e-05,
+      "loss": 0.0041,
+      "step": 26940
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.7505740740740745e-05,
+      "loss": 0.0062,
+      "step": 26950
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.750481481481482e-05,
+      "loss": 0.0078,
+      "step": 26960
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.7503888888888896e-05,
+      "loss": 0.0056,
+      "step": 26970
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.7502962962962964e-05,
+      "loss": 0.0035,
+      "step": 26980
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.750203703703704e-05,
+      "loss": 0.0054,
+      "step": 26990
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.7501111111111114e-05,
+      "loss": 0.0047,
+      "step": 27000
+    },
+    {
+      "epoch": 5.0,
+      "eval_loss": 0.0124629195779562,
+      "eval_runtime": 153.3534,
+      "eval_samples_per_second": 26.018,
+      "eval_steps_per_second": 3.254,
+      "step": 27000
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.750018518518519e-05,
+      "loss": 0.0027,
+      "step": 27010
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.749925925925926e-05,
+      "loss": 0.0098,
+      "step": 27020
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.749833333333333e-05,
+      "loss": 0.0045,
+      "step": 27030
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.749740740740741e-05,
+      "loss": 0.0033,
+      "step": 27040
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.749648148148149e-05,
+      "loss": 0.0049,
+      "step": 27050
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.749555555555556e-05,
+      "loss": 0.0051,
+      "step": 27060
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.749462962962963e-05,
+      "loss": 0.0054,
+      "step": 27070
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.749370370370371e-05,
+      "loss": 0.0055,
+      "step": 27080
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.749277777777778e-05,
+      "loss": 0.0039,
+      "step": 27090
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.749185185185185e-05,
+      "loss": 0.0047,
+      "step": 27100
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.7490925925925926e-05,
+      "loss": 0.0094,
+      "step": 27110
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.749e-05,
+      "loss": 0.0074,
+      "step": 27120
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.7489074074074076e-05,
+      "loss": 0.0084,
+      "step": 27130
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.748814814814815e-05,
+      "loss": 0.003,
+      "step": 27140
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.748722222222222e-05,
+      "loss": 0.0085,
+      "step": 27150
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.74862962962963e-05,
+      "loss": 0.0117,
+      "step": 27160
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.7485370370370376e-05,
+      "loss": 0.0052,
+      "step": 27170
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.748444444444445e-05,
+      "loss": 0.0039,
+      "step": 27180
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.748351851851852e-05,
+      "loss": 0.0073,
+      "step": 27190
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.7482592592592595e-05,
+      "loss": 0.0051,
+      "step": 27200
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.748166666666667e-05,
+      "loss": 0.0042,
+      "step": 27210
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.7480740740740745e-05,
+      "loss": 0.0074,
+      "step": 27220
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.747981481481481e-05,
+      "loss": 0.0056,
+      "step": 27230
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.747888888888889e-05,
+      "loss": 0.0071,
+      "step": 27240
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.747796296296296e-05,
+      "loss": 0.0058,
+      "step": 27250
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.747703703703704e-05,
+      "loss": 0.005,
+      "step": 27260
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.747611111111111e-05,
+      "loss": 0.0024,
+      "step": 27270
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.747518518518519e-05,
+      "loss": 0.0033,
+      "step": 27280
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.7474259259259263e-05,
+      "loss": 0.0054,
+      "step": 27290
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.747333333333334e-05,
+      "loss": 0.0051,
+      "step": 27300
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.747240740740741e-05,
+      "loss": 0.0124,
+      "step": 27310
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.747148148148148e-05,
+      "loss": 0.0067,
+      "step": 27320
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.747055555555556e-05,
+      "loss": 0.0049,
+      "step": 27330
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.746962962962963e-05,
+      "loss": 0.0042,
+      "step": 27340
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.746870370370371e-05,
+      "loss": 0.0048,
+      "step": 27350
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.7467777777777775e-05,
+      "loss": 0.0077,
+      "step": 27360
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.746685185185186e-05,
+      "loss": 0.0045,
+      "step": 27370
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.746592592592593e-05,
+      "loss": 0.0054,
+      "step": 27380
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.746500000000001e-05,
+      "loss": 0.0048,
+      "step": 27390
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.7464074074074075e-05,
+      "loss": 0.0051,
+      "step": 27400
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.746314814814815e-05,
+      "loss": 0.0042,
+      "step": 27410
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.7462222222222226e-05,
+      "loss": 0.0069,
+      "step": 27420
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.74612962962963e-05,
+      "loss": 0.0072,
+      "step": 27430
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.746037037037037e-05,
+      "loss": 0.0039,
+      "step": 27440
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.7459444444444444e-05,
+      "loss": 0.0074,
+      "step": 27450
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.745851851851852e-05,
+      "loss": 0.0053,
+      "step": 27460
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.7457592592592594e-05,
+      "loss": 0.0078,
+      "step": 27470
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.745666666666667e-05,
+      "loss": 0.004,
+      "step": 27480
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.7455740740740744e-05,
+      "loss": 0.0074,
+      "step": 27490
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.745481481481482e-05,
+      "loss": 0.0066,
+      "step": 27500
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.7453888888888894e-05,
+      "loss": 0.0046,
+      "step": 27510
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.745296296296296e-05,
+      "loss": 0.0054,
+      "step": 27520
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.745203703703704e-05,
+      "loss": 0.0038,
+      "step": 27530
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.745111111111111e-05,
+      "loss": 0.0037,
+      "step": 27540
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.745018518518519e-05,
+      "loss": 0.0086,
+      "step": 27550
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.744925925925926e-05,
+      "loss": 0.0051,
+      "step": 27560
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.744833333333333e-05,
+      "loss": 0.0082,
+      "step": 27570
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.7447407407407406e-05,
+      "loss": 0.0095,
+      "step": 27580
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.744648148148149e-05,
+      "loss": 0.0038,
+      "step": 27590
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.744555555555556e-05,
+      "loss": 0.0046,
+      "step": 27600
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.744462962962963e-05,
+      "loss": 0.0062,
+      "step": 27610
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 4.7443703703703706e-05,
+      "loss": 0.0042,
+      "step": 27620
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.744277777777778e-05,
+      "loss": 0.0047,
+      "step": 27630
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.7441851851851856e-05,
+      "loss": 0.0052,
+      "step": 27640
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.7440925925925925e-05,
+      "loss": 0.0093,
+      "step": 27650
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.744e-05,
+      "loss": 0.0047,
+      "step": 27660
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.7439074074074075e-05,
+      "loss": 0.0072,
+      "step": 27670
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.743814814814815e-05,
+      "loss": 0.0043,
+      "step": 27680
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.7437222222222225e-05,
+      "loss": 0.0063,
+      "step": 27690
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.74362962962963e-05,
+      "loss": 0.005,
+      "step": 27700
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.7435370370370375e-05,
+      "loss": 0.0044,
+      "step": 27710
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.743444444444445e-05,
+      "loss": 0.0049,
+      "step": 27720
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.743351851851852e-05,
+      "loss": 0.0072,
+      "step": 27730
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.7432592592592593e-05,
+      "loss": 0.0052,
+      "step": 27740
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.743166666666667e-05,
+      "loss": 0.005,
+      "step": 27750
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.7430740740740744e-05,
+      "loss": 0.0069,
+      "step": 27760
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.742981481481482e-05,
+      "loss": 0.0057,
+      "step": 27770
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.742888888888889e-05,
+      "loss": 0.0063,
+      "step": 27780
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.742796296296296e-05,
+      "loss": 0.0058,
+      "step": 27790
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.7427037037037044e-05,
+      "loss": 0.0045,
+      "step": 27800
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.742611111111112e-05,
+      "loss": 0.0039,
+      "step": 27810
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.742518518518519e-05,
+      "loss": 0.0079,
+      "step": 27820
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.742425925925926e-05,
+      "loss": 0.0045,
+      "step": 27830
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.742333333333334e-05,
+      "loss": 0.0075,
+      "step": 27840
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.742240740740741e-05,
+      "loss": 0.0084,
+      "step": 27850
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.742148148148148e-05,
+      "loss": 0.0066,
+      "step": 27860
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.7420555555555556e-05,
+      "loss": 0.0066,
+      "step": 27870
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.741962962962963e-05,
+      "loss": 0.0073,
+      "step": 27880
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 4.7418703703703706e-05,
+      "loss": 0.0054,
+      "step": 27890
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.741777777777778e-05,
+      "loss": 0.0057,
+      "step": 27900
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.7416851851851856e-05,
+      "loss": 0.0047,
+      "step": 27910
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.741592592592593e-05,
+      "loss": 0.0061,
+      "step": 27920
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.7415000000000006e-05,
+      "loss": 0.0045,
+      "step": 27930
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.7414074074074074e-05,
+      "loss": 0.0063,
+      "step": 27940
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.741314814814815e-05,
+      "loss": 0.0042,
+      "step": 27950
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.7412222222222224e-05,
+      "loss": 0.0062,
+      "step": 27960
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.74112962962963e-05,
+      "loss": 0.0053,
+      "step": 27970
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.7410370370370374e-05,
+      "loss": 0.0029,
+      "step": 27980
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.740944444444444e-05,
+      "loss": 0.006,
+      "step": 27990
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.740851851851852e-05,
+      "loss": 0.0044,
+      "step": 28000
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.740759259259259e-05,
+      "loss": 0.0035,
+      "step": 28010
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.7406666666666675e-05,
+      "loss": 0.0064,
+      "step": 28020
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.740574074074074e-05,
+      "loss": 0.0046,
+      "step": 28030
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.740481481481482e-05,
+      "loss": 0.0056,
+      "step": 28040
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.740388888888889e-05,
+      "loss": 0.0044,
+      "step": 28050
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.740296296296297e-05,
+      "loss": 0.0076,
+      "step": 28060
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.7402037037037036e-05,
+      "loss": 0.0057,
+      "step": 28070
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.740111111111111e-05,
+      "loss": 0.0077,
+      "step": 28080
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.7400185185185187e-05,
+      "loss": 0.0062,
+      "step": 28090
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.739925925925926e-05,
+      "loss": 0.0052,
+      "step": 28100
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.739833333333334e-05,
+      "loss": 0.005,
+      "step": 28110
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.7397407407407405e-05,
+      "loss": 0.0042,
+      "step": 28120
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.739648148148149e-05,
+      "loss": 0.0057,
+      "step": 28130
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.739555555555556e-05,
+      "loss": 0.0041,
+      "step": 28140
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.739462962962963e-05,
+      "loss": 0.0069,
+      "step": 28150
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.7393703703703705e-05,
+      "loss": 0.0062,
+      "step": 28160
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.739277777777778e-05,
+      "loss": 0.0052,
+      "step": 28170
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.7391851851851855e-05,
+      "loss": 0.0046,
+      "step": 28180
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.739092592592593e-05,
+      "loss": 0.0065,
+      "step": 28190
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.739e-05,
+      "loss": 0.0047,
+      "step": 28200
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.7389074074074074e-05,
+      "loss": 0.0058,
+      "step": 28210
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.738814814814815e-05,
+      "loss": 0.0066,
+      "step": 28220
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.7387222222222224e-05,
+      "loss": 0.0069,
+      "step": 28230
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.73862962962963e-05,
+      "loss": 0.0053,
+      "step": 28240
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.7385370370370374e-05,
+      "loss": 0.0063,
+      "step": 28250
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.738444444444445e-05,
+      "loss": 0.0034,
+      "step": 28260
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.7383518518518524e-05,
+      "loss": 0.0082,
+      "step": 28270
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.738259259259259e-05,
+      "loss": 0.0058,
+      "step": 28280
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.738166666666667e-05,
+      "loss": 0.0063,
+      "step": 28290
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.738074074074074e-05,
+      "loss": 0.0032,
+      "step": 28300
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.737981481481482e-05,
+      "loss": 0.0054,
+      "step": 28310
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.737888888888889e-05,
+      "loss": 0.0038,
+      "step": 28320
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.737796296296296e-05,
+      "loss": 0.0034,
+      "step": 28330
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.737703703703704e-05,
+      "loss": 0.0056,
+      "step": 28340
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.737611111111112e-05,
+      "loss": 0.0048,
+      "step": 28350
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.7375185185185186e-05,
+      "loss": 0.005,
+      "step": 28360
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.737425925925926e-05,
+      "loss": 0.0044,
+      "step": 28370
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.7373333333333336e-05,
+      "loss": 0.0071,
+      "step": 28380
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.737240740740741e-05,
+      "loss": 0.0051,
+      "step": 28390
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.7371481481481486e-05,
+      "loss": 0.0062,
+      "step": 28400
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.7370555555555554e-05,
+      "loss": 0.0064,
+      "step": 28410
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.736962962962963e-05,
+      "loss": 0.0055,
+      "step": 28420
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.7368703703703705e-05,
+      "loss": 0.0066,
+      "step": 28430
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.736777777777778e-05,
+      "loss": 0.0045,
+      "step": 28440
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.7366851851851855e-05,
+      "loss": 0.0044,
+      "step": 28450
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.736592592592593e-05,
+      "loss": 0.006,
+      "step": 28460
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.7365000000000005e-05,
+      "loss": 0.0049,
+      "step": 28470
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.736407407407408e-05,
+      "loss": 0.0068,
+      "step": 28480
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.736314814814815e-05,
+      "loss": 0.0045,
+      "step": 28490
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.736222222222222e-05,
+      "loss": 0.0056,
+      "step": 28500
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.73612962962963e-05,
+      "loss": 0.0078,
+      "step": 28510
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.736037037037037e-05,
+      "loss": 0.0058,
+      "step": 28520
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.735944444444445e-05,
+      "loss": 0.0056,
+      "step": 28530
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.7358518518518517e-05,
+      "loss": 0.0045,
+      "step": 28540
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.735759259259259e-05,
+      "loss": 0.0042,
+      "step": 28550
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.7356666666666673e-05,
+      "loss": 0.0053,
+      "step": 28560
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.735574074074074e-05,
+      "loss": 0.0054,
+      "step": 28570
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.735481481481482e-05,
+      "loss": 0.0053,
+      "step": 28580
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.735388888888889e-05,
+      "loss": 0.0045,
+      "step": 28590
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.735305555555556e-05,
+      "loss": 0.0047,
+      "step": 28600
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.735212962962963e-05,
+      "loss": 0.0063,
+      "step": 28610
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.735120370370371e-05,
+      "loss": 0.0068,
+      "step": 28620
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.735027777777778e-05,
+      "loss": 0.0039,
+      "step": 28630
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.734935185185185e-05,
+      "loss": 0.0041,
+      "step": 28640
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.7348425925925925e-05,
+      "loss": 0.0055,
+      "step": 28650
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.73475e-05,
+      "loss": 0.0051,
+      "step": 28660
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.7346574074074075e-05,
+      "loss": 0.0078,
+      "step": 28670
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.734564814814815e-05,
+      "loss": 0.0075,
+      "step": 28680
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.7344722222222225e-05,
+      "loss": 0.0061,
+      "step": 28690
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 4.73437962962963e-05,
+      "loss": 0.0077,
+      "step": 28700
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.7342870370370376e-05,
+      "loss": 0.0048,
+      "step": 28710
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.7341944444444444e-05,
+      "loss": 0.0037,
+      "step": 28720
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.734101851851852e-05,
+      "loss": 0.0047,
+      "step": 28730
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.7340092592592594e-05,
+      "loss": 0.0053,
+      "step": 28740
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.733916666666667e-05,
+      "loss": 0.0042,
+      "step": 28750
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.7338240740740744e-05,
+      "loss": 0.0053,
+      "step": 28760
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.733731481481481e-05,
+      "loss": 0.0083,
+      "step": 28770
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.733638888888889e-05,
+      "loss": 0.0063,
+      "step": 28780
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.733546296296297e-05,
+      "loss": 0.0085,
+      "step": 28790
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.7334537037037044e-05,
+      "loss": 0.006,
+      "step": 28800
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.733361111111111e-05,
+      "loss": 0.0048,
+      "step": 28810
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.733268518518519e-05,
+      "loss": 0.0069,
+      "step": 28820
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.733175925925926e-05,
+      "loss": 0.0049,
+      "step": 28830
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.733083333333334e-05,
+      "loss": 0.0041,
+      "step": 28840
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.7329907407407406e-05,
+      "loss": 0.0051,
+      "step": 28850
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.732898148148148e-05,
+      "loss": 0.0035,
+      "step": 28860
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.7328055555555556e-05,
+      "loss": 0.0052,
+      "step": 28870
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.732712962962963e-05,
+      "loss": 0.0026,
+      "step": 28880
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.7326203703703706e-05,
+      "loss": 0.0062,
+      "step": 28890
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.732527777777778e-05,
+      "loss": 0.0053,
+      "step": 28900
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.7324351851851856e-05,
+      "loss": 0.0027,
+      "step": 28910
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.732342592592593e-05,
+      "loss": 0.0051,
+      "step": 28920
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.73225e-05,
+      "loss": 0.0053,
+      "step": 28930
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.7321574074074075e-05,
+      "loss": 0.0086,
+      "step": 28940
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.732064814814815e-05,
+      "loss": 0.0047,
+      "step": 28950
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.7319722222222225e-05,
+      "loss": 0.0093,
+      "step": 28960
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 4.73187962962963e-05,
+      "loss": 0.0036,
+      "step": 28970
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.731787037037037e-05,
+      "loss": 0.0053,
+      "step": 28980
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.731694444444444e-05,
+      "loss": 0.0055,
+      "step": 28990
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.7316018518518525e-05,
+      "loss": 0.0047,
+      "step": 29000
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.73150925925926e-05,
+      "loss": 0.0044,
+      "step": 29010
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.731416666666667e-05,
+      "loss": 0.0071,
+      "step": 29020
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.7313240740740743e-05,
+      "loss": 0.0054,
+      "step": 29030
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.731231481481482e-05,
+      "loss": 0.0033,
+      "step": 29040
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.7311388888888894e-05,
+      "loss": 0.0073,
+      "step": 29050
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.731046296296296e-05,
+      "loss": 0.0047,
+      "step": 29060
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 4.730953703703704e-05,
+      "loss": 0.0053,
+      "step": 29070
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.730861111111111e-05,
+      "loss": 0.0043,
+      "step": 29080
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.730768518518519e-05,
+      "loss": 0.0057,
+      "step": 29090
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.730675925925926e-05,
+      "loss": 0.0057,
+      "step": 29100
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.730583333333334e-05,
+      "loss": 0.0055,
+      "step": 29110
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.730490740740741e-05,
+      "loss": 0.0053,
+      "step": 29120
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.730398148148149e-05,
+      "loss": 0.0077,
+      "step": 29130
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.7303055555555555e-05,
+      "loss": 0.0041,
+      "step": 29140
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.730212962962963e-05,
+      "loss": 0.0051,
+      "step": 29150
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.7301203703703706e-05,
+      "loss": 0.0083,
+      "step": 29160
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.730027777777778e-05,
+      "loss": 0.0055,
+      "step": 29170
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.7299351851851856e-05,
+      "loss": 0.0039,
+      "step": 29180
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.7298425925925924e-05,
+      "loss": 0.0046,
+      "step": 29190
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.72975e-05,
+      "loss": 0.0062,
+      "step": 29200
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.7296574074074074e-05,
+      "loss": 0.0048,
+      "step": 29210
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.7295648148148156e-05,
+      "loss": 0.004,
+      "step": 29220
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.7294722222222224e-05,
+      "loss": 0.0067,
+      "step": 29230
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.72937962962963e-05,
+      "loss": 0.0046,
+      "step": 29240
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.7292870370370374e-05,
+      "loss": 0.0049,
+      "step": 29250
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.729194444444445e-05,
+      "loss": 0.0043,
+      "step": 29260
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.729101851851852e-05,
+      "loss": 0.0052,
+      "step": 29270
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.729009259259259e-05,
+      "loss": 0.0072,
+      "step": 29280
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.728916666666667e-05,
+      "loss": 0.0052,
+      "step": 29290
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.728824074074074e-05,
+      "loss": 0.0072,
+      "step": 29300
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.728731481481482e-05,
+      "loss": 0.0032,
+      "step": 29310
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.7286388888888886e-05,
+      "loss": 0.0056,
+      "step": 29320
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.728546296296297e-05,
+      "loss": 0.0053,
+      "step": 29330
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.728453703703704e-05,
+      "loss": 0.004,
+      "step": 29340
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.728361111111111e-05,
+      "loss": 0.0068,
+      "step": 29350
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.7282685185185186e-05,
+      "loss": 0.0058,
+      "step": 29360
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.728175925925926e-05,
+      "loss": 0.0043,
+      "step": 29370
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.7280833333333336e-05,
+      "loss": 0.0046,
+      "step": 29380
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.727990740740741e-05,
+      "loss": 0.0066,
+      "step": 29390
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.727898148148148e-05,
+      "loss": 0.0044,
+      "step": 29400
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.7278055555555555e-05,
+      "loss": 0.0056,
+      "step": 29410
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.727712962962963e-05,
+      "loss": 0.0054,
+      "step": 29420
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.727620370370371e-05,
+      "loss": 0.0073,
+      "step": 29430
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.727527777777778e-05,
+      "loss": 0.0047,
+      "step": 29440
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.7274351851851855e-05,
+      "loss": 0.0062,
+      "step": 29450
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.727342592592593e-05,
+      "loss": 0.0038,
+      "step": 29460
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.7272500000000005e-05,
+      "loss": 0.0049,
+      "step": 29470
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.7271574074074073e-05,
+      "loss": 0.0077,
+      "step": 29480
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.727064814814815e-05,
+      "loss": 0.0051,
+      "step": 29490
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.7269722222222224e-05,
+      "loss": 0.0053,
+      "step": 29500
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.726888888888889e-05,
+      "loss": 0.0057,
+      "step": 29510
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.7267962962962964e-05,
+      "loss": 0.006,
+      "step": 29520
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.726703703703704e-05,
+      "loss": 0.0036,
+      "step": 29530
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.7266111111111114e-05,
+      "loss": 0.0047,
+      "step": 29540
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.726518518518519e-05,
+      "loss": 0.0044,
+      "step": 29550
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.7264259259259264e-05,
+      "loss": 0.0082,
+      "step": 29560
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.726333333333334e-05,
+      "loss": 0.0066,
+      "step": 29570
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.7262407407407414e-05,
+      "loss": 0.0051,
+      "step": 29580
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.726148148148148e-05,
+      "loss": 0.005,
+      "step": 29590
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.726055555555556e-05,
+      "loss": 0.0044,
+      "step": 29600
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.725962962962963e-05,
+      "loss": 0.0059,
+      "step": 29610
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.725870370370371e-05,
+      "loss": 0.004,
+      "step": 29620
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.7257777777777776e-05,
+      "loss": 0.0047,
+      "step": 29630
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.725685185185185e-05,
+      "loss": 0.0056,
+      "step": 29640
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.7255925925925926e-05,
+      "loss": 0.005,
+      "step": 29650
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.725500000000001e-05,
+      "loss": 0.0044,
+      "step": 29660
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.7254074074074076e-05,
+      "loss": 0.004,
+      "step": 29670
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.725314814814815e-05,
+      "loss": 0.0033,
+      "step": 29680
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.7252222222222226e-05,
+      "loss": 0.0044,
+      "step": 29690
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.72512962962963e-05,
+      "loss": 0.0049,
+      "step": 29700
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.7250370370370376e-05,
+      "loss": 0.0049,
+      "step": 29710
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.7249444444444444e-05,
+      "loss": 0.0064,
+      "step": 29720
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.724851851851852e-05,
+      "loss": 0.0066,
+      "step": 29730
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.7247592592592594e-05,
+      "loss": 0.0035,
+      "step": 29740
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.724666666666667e-05,
+      "loss": 0.0065,
+      "step": 29750
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.724574074074074e-05,
+      "loss": 0.0048,
+      "step": 29760
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.724481481481482e-05,
+      "loss": 0.0051,
+      "step": 29770
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.7243888888888895e-05,
+      "loss": 0.0045,
+      "step": 29780
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.724296296296297e-05,
+      "loss": 0.0086,
+      "step": 29790
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.724203703703704e-05,
+      "loss": 0.0047,
+      "step": 29800
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.724111111111111e-05,
+      "loss": 0.0084,
+      "step": 29810
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.724018518518519e-05,
+      "loss": 0.0072,
+      "step": 29820
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.723925925925926e-05,
+      "loss": 0.0069,
+      "step": 29830
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.723833333333333e-05,
+      "loss": 0.0064,
+      "step": 29840
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.7237407407407406e-05,
+      "loss": 0.0037,
+      "step": 29850
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.723648148148148e-05,
+      "loss": 0.0031,
+      "step": 29860
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.7235555555555557e-05,
+      "loss": 0.0056,
+      "step": 29870
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.723462962962963e-05,
+      "loss": 0.0077,
+      "step": 29880
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.723370370370371e-05,
+      "loss": 0.0061,
+      "step": 29890
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.723277777777778e-05,
+      "loss": 0.006,
+      "step": 29900
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.723185185185186e-05,
+      "loss": 0.0061,
+      "step": 29910
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.723092592592593e-05,
+      "loss": 0.0033,
+      "step": 29920
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.723e-05,
+      "loss": 0.0032,
+      "step": 29930
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 4.7229074074074075e-05,
+      "loss": 0.0052,
+      "step": 29940
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.722814814814815e-05,
+      "loss": 0.005,
+      "step": 29950
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.7227222222222225e-05,
+      "loss": 0.0088,
+      "step": 29960
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.7226296296296294e-05,
+      "loss": 0.0071,
+      "step": 29970
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.7225370370370375e-05,
+      "loss": 0.0081,
+      "step": 29980
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.722444444444445e-05,
+      "loss": 0.0067,
+      "step": 29990
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.7223518518518526e-05,
+      "loss": 0.0065,
+      "step": 30000
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.7222592592592594e-05,
+      "loss": 0.0091,
+      "step": 30010
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.722166666666667e-05,
+      "loss": 0.0052,
+      "step": 30020
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.7220740740740744e-05,
+      "loss": 0.0041,
+      "step": 30030
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.721981481481482e-05,
+      "loss": 0.0058,
+      "step": 30040
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 4.721888888888889e-05,
+      "loss": 0.0027,
+      "step": 30050
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.721796296296296e-05,
+      "loss": 0.0061,
+      "step": 30060
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.721703703703704e-05,
+      "loss": 0.004,
+      "step": 30070
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.721611111111111e-05,
+      "loss": 0.006,
+      "step": 30080
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.721518518518519e-05,
+      "loss": 0.0074,
+      "step": 30090
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.721425925925926e-05,
+      "loss": 0.0028,
+      "step": 30100
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.721333333333334e-05,
+      "loss": 0.0078,
+      "step": 30110
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.721240740740741e-05,
+      "loss": 0.0035,
+      "step": 30120
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.721148148148149e-05,
+      "loss": 0.0061,
+      "step": 30130
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.7210555555555556e-05,
+      "loss": 0.005,
+      "step": 30140
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 4.720962962962963e-05,
+      "loss": 0.0044,
+      "step": 30150
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.7208703703703706e-05,
+      "loss": 0.0037,
+      "step": 30160
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.720777777777778e-05,
+      "loss": 0.0091,
+      "step": 30170
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.720685185185185e-05,
+      "loss": 0.0056,
+      "step": 30180
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.7205925925925924e-05,
+      "loss": 0.0044,
+      "step": 30190
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.7205000000000006e-05,
+      "loss": 0.0063,
+      "step": 30200
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.720407407407408e-05,
+      "loss": 0.0051,
+      "step": 30210
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.720314814814815e-05,
+      "loss": 0.0055,
+      "step": 30220
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.7202222222222225e-05,
+      "loss": 0.0075,
+      "step": 30230
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.72012962962963e-05,
+      "loss": 0.0046,
+      "step": 30240
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.7200370370370375e-05,
+      "loss": 0.0035,
+      "step": 30250
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 4.719944444444444e-05,
+      "loss": 0.0068,
+      "step": 30260
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.719851851851852e-05,
+      "loss": 0.007,
+      "step": 30270
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.719759259259259e-05,
+      "loss": 0.0037,
+      "step": 30280
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.719666666666667e-05,
+      "loss": 0.0039,
+      "step": 30290
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.719574074074074e-05,
+      "loss": 0.0033,
+      "step": 30300
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.719481481481482e-05,
+      "loss": 0.0065,
+      "step": 30310
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 4.7193888888888893e-05,
+      "loss": 0.0072,
+      "step": 30320
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.719296296296297e-05,
+      "loss": 0.0052,
+      "step": 30330
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.7192037037037044e-05,
+      "loss": 0.0056,
+      "step": 30340
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.719111111111111e-05,
+      "loss": 0.0063,
+      "step": 30350
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.719018518518519e-05,
+      "loss": 0.0036,
+      "step": 30360
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 4.718925925925926e-05,
+      "loss": 0.0117,
+      "step": 30370
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.718833333333334e-05,
+      "loss": 0.0038,
+      "step": 30380
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.7187407407407405e-05,
+      "loss": 0.0052,
+      "step": 30390
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.718648148148148e-05,
+      "loss": 0.0081,
+      "step": 30400
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.7185555555555555e-05,
+      "loss": 0.0055,
+      "step": 30410
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 4.718462962962964e-05,
+      "loss": 0.0067,
+      "step": 30420
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.7183703703703705e-05,
+      "loss": 0.0088,
+      "step": 30430
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.718277777777778e-05,
+      "loss": 0.0045,
+      "step": 30440
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.7181851851851856e-05,
+      "loss": 0.0057,
+      "step": 30450
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.718092592592593e-05,
+      "loss": 0.0048,
+      "step": 30460
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.718e-05,
+      "loss": 0.0058,
+      "step": 30470
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 4.7179074074074074e-05,
+      "loss": 0.0055,
+      "step": 30480
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.717814814814815e-05,
+      "loss": 0.005,
+      "step": 30490
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.7177222222222224e-05,
+      "loss": 0.0062,
+      "step": 30500
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.71762962962963e-05,
+      "loss": 0.0075,
+      "step": 30510
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.7175370370370374e-05,
+      "loss": 0.0043,
+      "step": 30520
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 4.717444444444445e-05,
+      "loss": 0.004,
+      "step": 30530
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.7173518518518524e-05,
+      "loss": 0.0044,
+      "step": 30540
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.71725925925926e-05,
+      "loss": 0.0029,
+      "step": 30550
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.717166666666667e-05,
+      "loss": 0.003,
+      "step": 30560
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.717074074074074e-05,
+      "loss": 0.006,
+      "step": 30570
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.716981481481482e-05,
+      "loss": 0.0051,
+      "step": 30580
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 4.716888888888889e-05,
+      "loss": 0.0065,
+      "step": 30590
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.716796296296296e-05,
+      "loss": 0.0062,
+      "step": 30600
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.7167037037037036e-05,
+      "loss": 0.0034,
+      "step": 30610
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.716611111111111e-05,
+      "loss": 0.0035,
+      "step": 30620
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.716518518518519e-05,
+      "loss": 0.0074,
+      "step": 30630
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 4.716425925925926e-05,
+      "loss": 0.0038,
+      "step": 30640
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.7163333333333336e-05,
+      "loss": 0.0038,
+      "step": 30650
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.716240740740741e-05,
+      "loss": 0.0041,
+      "step": 30660
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.7161481481481486e-05,
+      "loss": 0.004,
+      "step": 30670
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.7160555555555555e-05,
+      "loss": 0.0032,
+      "step": 30680
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 4.715962962962963e-05,
+      "loss": 0.0058,
+      "step": 30690
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.7158703703703705e-05,
+      "loss": 0.0062,
+      "step": 30700
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.715777777777778e-05,
+      "loss": 0.0064,
+      "step": 30710
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.7156851851851855e-05,
+      "loss": 0.0042,
+      "step": 30720
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.715592592592592e-05,
+      "loss": 0.0036,
+      "step": 30730
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.7155000000000005e-05,
+      "loss": 0.0054,
+      "step": 30740
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 4.715407407407408e-05,
+      "loss": 0.0054,
+      "step": 30750
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.7153148148148155e-05,
+      "loss": 0.0059,
+      "step": 30760
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.7152222222222223e-05,
+      "loss": 0.0051,
+      "step": 30770
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.71512962962963e-05,
+      "loss": 0.0068,
+      "step": 30780
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.7150370370370374e-05,
+      "loss": 0.0049,
+      "step": 30790
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 4.714944444444445e-05,
+      "loss": 0.0066,
+      "step": 30800
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.714851851851852e-05,
+      "loss": 0.0054,
+      "step": 30810
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.714759259259259e-05,
+      "loss": 0.0062,
+      "step": 30820
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.714666666666667e-05,
+      "loss": 0.0069,
+      "step": 30830
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.714574074074074e-05,
+      "loss": 0.0059,
+      "step": 30840
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.714481481481482e-05,
+      "loss": 0.0097,
+      "step": 30850
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 4.714388888888889e-05,
+      "loss": 0.0054,
+      "step": 30860
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.714296296296297e-05,
+      "loss": 0.0057,
+      "step": 30870
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.714203703703704e-05,
+      "loss": 0.005,
+      "step": 30880
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.714111111111111e-05,
+      "loss": 0.0062,
+      "step": 30890
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.7140185185185186e-05,
+      "loss": 0.0068,
+      "step": 30900
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 4.713925925925926e-05,
+      "loss": 0.0063,
+      "step": 30910
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.7138333333333336e-05,
+      "loss": 0.0079,
+      "step": 30920
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.713740740740741e-05,
+      "loss": 0.0059,
+      "step": 30930
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.713648148148148e-05,
+      "loss": 0.0055,
+      "step": 30940
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.713555555555556e-05,
+      "loss": 0.005,
+      "step": 30950
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 4.7134629629629636e-05,
+      "loss": 0.0054,
+      "step": 30960
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.713370370370371e-05,
+      "loss": 0.0036,
+      "step": 30970
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.713277777777778e-05,
+      "loss": 0.0039,
+      "step": 30980
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.7131851851851854e-05,
+      "loss": 0.0051,
+      "step": 30990
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.713092592592593e-05,
+      "loss": 0.0037,
+      "step": 31000
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.7130000000000004e-05,
+      "loss": 0.0028,
+      "step": 31010
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 4.712907407407407e-05,
+      "loss": 0.0054,
+      "step": 31020
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.712814814814815e-05,
+      "loss": 0.0074,
+      "step": 31030
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.712722222222222e-05,
+      "loss": 0.0057,
+      "step": 31040
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.71262962962963e-05,
+      "loss": 0.0049,
+      "step": 31050
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.712537037037037e-05,
+      "loss": 0.004,
+      "step": 31060
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 4.712444444444445e-05,
+      "loss": 0.0076,
+      "step": 31070
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.712351851851852e-05,
+      "loss": 0.0057,
+      "step": 31080
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.71225925925926e-05,
+      "loss": 0.0064,
+      "step": 31090
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.7121666666666666e-05,
+      "loss": 0.0058,
+      "step": 31100
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.712074074074074e-05,
+      "loss": 0.0047,
+      "step": 31110
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.7119814814814817e-05,
+      "loss": 0.0077,
+      "step": 31120
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 4.711888888888889e-05,
+      "loss": 0.0057,
+      "step": 31130
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.711796296296297e-05,
+      "loss": 0.0047,
+      "step": 31140
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.7117037037037035e-05,
+      "loss": 0.0036,
+      "step": 31150
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.711611111111111e-05,
+      "loss": 0.004,
+      "step": 31160
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.711518518518519e-05,
+      "loss": 0.007,
+      "step": 31170
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 4.711425925925927e-05,
+      "loss": 0.0063,
+      "step": 31180
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.7113333333333335e-05,
+      "loss": 0.0039,
+      "step": 31190
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.711240740740741e-05,
+      "loss": 0.0048,
+      "step": 31200
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.7111481481481485e-05,
+      "loss": 0.0049,
+      "step": 31210
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.711055555555556e-05,
+      "loss": 0.0083,
+      "step": 31220
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 4.710962962962963e-05,
+      "loss": 0.0044,
+      "step": 31230
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.7108703703703704e-05,
+      "loss": 0.0071,
+      "step": 31240
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.710777777777778e-05,
+      "loss": 0.0066,
+      "step": 31250
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.7106851851851854e-05,
+      "loss": 0.0043,
+      "step": 31260
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.710592592592593e-05,
+      "loss": 0.0045,
+      "step": 31270
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.7105000000000004e-05,
+      "loss": 0.0102,
+      "step": 31280
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 4.710407407407408e-05,
+      "loss": 0.0061,
+      "step": 31290
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.7103148148148154e-05,
+      "loss": 0.0043,
+      "step": 31300
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.710222222222222e-05,
+      "loss": 0.0066,
+      "step": 31310
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.71012962962963e-05,
+      "loss": 0.0045,
+      "step": 31320
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.710037037037037e-05,
+      "loss": 0.0037,
+      "step": 31330
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 4.709944444444445e-05,
+      "loss": 0.0042,
+      "step": 31340
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.709851851851852e-05,
+      "loss": 0.0048,
+      "step": 31350
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.709759259259259e-05,
+      "loss": 0.0063,
+      "step": 31360
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.7096666666666666e-05,
+      "loss": 0.0051,
+      "step": 31370
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.709574074074074e-05,
+      "loss": 0.0045,
+      "step": 31380
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.709481481481482e-05,
+      "loss": 0.006,
+      "step": 31390
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 4.709388888888889e-05,
+      "loss": 0.0051,
+      "step": 31400
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.7092962962962966e-05,
+      "loss": 0.0031,
+      "step": 31410
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.709203703703704e-05,
+      "loss": 0.0054,
+      "step": 31420
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.7091111111111116e-05,
+      "loss": 0.0041,
+      "step": 31430
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.7090185185185184e-05,
+      "loss": 0.0063,
+      "step": 31440
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.708925925925926e-05,
+      "loss": 0.0038,
+      "step": 31450
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.7088333333333335e-05,
+      "loss": 0.0098,
+      "step": 31460
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.708740740740741e-05,
+      "loss": 0.0034,
+      "step": 31470
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.7086481481481485e-05,
+      "loss": 0.0057,
+      "step": 31480
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.708555555555556e-05,
+      "loss": 0.004,
+      "step": 31490
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.7084629629629635e-05,
+      "loss": 0.0058,
+      "step": 31500
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.708370370370371e-05,
+      "loss": 0.006,
+      "step": 31510
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.708277777777778e-05,
+      "loss": 0.0053,
+      "step": 31520
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.708185185185185e-05,
+      "loss": 0.005,
+      "step": 31530
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.708092592592593e-05,
+      "loss": 0.0057,
+      "step": 31540
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.708e-05,
+      "loss": 0.0054,
+      "step": 31550
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 4.707907407407408e-05,
+      "loss": 0.0056,
+      "step": 31560
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.7078148148148147e-05,
+      "loss": 0.0058,
+      "step": 31570
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.707722222222222e-05,
+      "loss": 0.0069,
+      "step": 31580
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.70762962962963e-05,
+      "loss": 0.0062,
+      "step": 31590
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.707537037037038e-05,
+      "loss": 0.0056,
+      "step": 31600
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 4.707444444444445e-05,
+      "loss": 0.0064,
+      "step": 31610
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.707351851851852e-05,
+      "loss": 0.0032,
+      "step": 31620
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.70725925925926e-05,
+      "loss": 0.0045,
+      "step": 31630
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.707166666666667e-05,
+      "loss": 0.0067,
+      "step": 31640
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.707074074074074e-05,
+      "loss": 0.0057,
+      "step": 31650
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.7069814814814815e-05,
+      "loss": 0.0047,
+      "step": 31660
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.706888888888889e-05,
+      "loss": 0.0081,
+      "step": 31670
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.7067962962962965e-05,
+      "loss": 0.0054,
+      "step": 31680
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.706703703703704e-05,
+      "loss": 0.0066,
+      "step": 31690
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.706611111111111e-05,
+      "loss": 0.0062,
+      "step": 31700
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.706518518518519e-05,
+      "loss": 0.0058,
+      "step": 31710
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 4.7064259259259266e-05,
+      "loss": 0.0045,
+      "step": 31720
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.7063333333333334e-05,
+      "loss": 0.005,
+      "step": 31730
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.706240740740741e-05,
+      "loss": 0.0036,
+      "step": 31740
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.7061481481481484e-05,
+      "loss": 0.0035,
+      "step": 31750
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.706055555555556e-05,
+      "loss": 0.005,
+      "step": 31760
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.7059629629629634e-05,
+      "loss": 0.0046,
+      "step": 31770
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.70587037037037e-05,
+      "loss": 0.0064,
+      "step": 31780
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.705777777777778e-05,
+      "loss": 0.0094,
+      "step": 31790
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.705685185185185e-05,
+      "loss": 0.0032,
+      "step": 31800
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.705592592592593e-05,
+      "loss": 0.0059,
+      "step": 31810
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.7055e-05,
+      "loss": 0.0072,
+      "step": 31820
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 4.705407407407408e-05,
+      "loss": 0.0034,
+      "step": 31830
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.705314814814815e-05,
+      "loss": 0.0041,
+      "step": 31840
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.705222222222223e-05,
+      "loss": 0.005,
+      "step": 31850
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.7051296296296296e-05,
+      "loss": 0.0049,
+      "step": 31860
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.705037037037037e-05,
+      "loss": 0.0086,
+      "step": 31870
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 4.7049444444444446e-05,
+      "loss": 0.0037,
+      "step": 31880
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.704851851851852e-05,
+      "loss": 0.0078,
+      "step": 31890
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.7047592592592596e-05,
+      "loss": 0.0108,
+      "step": 31900
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.7046666666666665e-05,
+      "loss": 0.004,
+      "step": 31910
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.704574074074074e-05,
+      "loss": 0.0052,
+      "step": 31920
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.704481481481482e-05,
+      "loss": 0.0054,
+      "step": 31930
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 4.704388888888889e-05,
+      "loss": 0.0038,
+      "step": 31940
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.7042962962962965e-05,
+      "loss": 0.0056,
+      "step": 31950
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.704203703703704e-05,
+      "loss": 0.0058,
+      "step": 31960
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.7041111111111115e-05,
+      "loss": 0.0053,
+      "step": 31970
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.704018518518519e-05,
+      "loss": 0.006,
+      "step": 31980
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 4.703925925925926e-05,
+      "loss": 0.0056,
+      "step": 31990
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.703833333333333e-05,
+      "loss": 0.005,
+      "step": 32000
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.703740740740741e-05,
+      "loss": 0.0034,
+      "step": 32010
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.7036481481481483e-05,
+      "loss": 0.0052,
+      "step": 32020
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.703555555555556e-05,
+      "loss": 0.01,
+      "step": 32030
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 4.7034629629629634e-05,
+      "loss": 0.0065,
+      "step": 32040
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.703370370370371e-05,
+      "loss": 0.0057,
+      "step": 32050
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.7032777777777784e-05,
+      "loss": 0.0042,
+      "step": 32060
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.703185185185185e-05,
+      "loss": 0.004,
+      "step": 32070
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.703092592592593e-05,
+      "loss": 0.0064,
+      "step": 32080
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.703e-05,
+      "loss": 0.0047,
+      "step": 32090
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.702907407407408e-05,
+      "loss": 0.007,
+      "step": 32100
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.702814814814815e-05,
+      "loss": 0.0031,
+      "step": 32110
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.702722222222222e-05,
+      "loss": 0.0065,
+      "step": 32120
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.7026296296296295e-05,
+      "loss": 0.0025,
+      "step": 32130
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.702537037037038e-05,
+      "loss": 0.004,
+      "step": 32140
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.7024444444444446e-05,
+      "loss": 0.0059,
+      "step": 32150
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.702351851851852e-05,
+      "loss": 0.004,
+      "step": 32160
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.7022592592592596e-05,
+      "loss": 0.0056,
+      "step": 32170
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.702166666666667e-05,
+      "loss": 0.008,
+      "step": 32180
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.7020740740740746e-05,
+      "loss": 0.0075,
+      "step": 32190
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.7019814814814814e-05,
+      "loss": 0.0034,
+      "step": 32200
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 4.701888888888889e-05,
+      "loss": 0.005,
+      "step": 32210
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.7017962962962964e-05,
+      "loss": 0.0039,
+      "step": 32220
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.701703703703704e-05,
+      "loss": 0.0105,
+      "step": 32230
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.701611111111111e-05,
+      "loss": 0.0054,
+      "step": 32240
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.701518518518519e-05,
+      "loss": 0.0063,
+      "step": 32250
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 4.7014259259259264e-05,
+      "loss": 0.0064,
+      "step": 32260
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.701333333333334e-05,
+      "loss": 0.0039,
+      "step": 32270
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.701240740740741e-05,
+      "loss": 0.005,
+      "step": 32280
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.701148148148148e-05,
+      "loss": 0.0052,
+      "step": 32290
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.701055555555556e-05,
+      "loss": 0.0048,
+      "step": 32300
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.700962962962963e-05,
+      "loss": 0.0048,
+      "step": 32310
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.700870370370371e-05,
+      "loss": 0.0053,
+      "step": 32320
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.7007777777777776e-05,
+      "loss": 0.0043,
+      "step": 32330
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.700685185185185e-05,
+      "loss": 0.0051,
+      "step": 32340
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.7005925925925926e-05,
+      "loss": 0.0055,
+      "step": 32350
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.7005e-05,
+      "loss": 0.0056,
+      "step": 32360
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.7004074074074076e-05,
+      "loss": 0.0053,
+      "step": 32370
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.700314814814815e-05,
+      "loss": 0.0023,
+      "step": 32380
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.7002222222222227e-05,
+      "loss": 0.0051,
+      "step": 32390
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.70012962962963e-05,
+      "loss": 0.0044,
+      "step": 32400
+    },
+    {
+      "epoch": 6.0,
+      "eval_loss": 0.013891405425965786,
+      "eval_runtime": 153.4566,
+      "eval_samples_per_second": 26.001,
+      "eval_steps_per_second": 3.252,
+      "step": 32400
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.700037037037037e-05,
+      "loss": 0.0024,
+      "step": 32410
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 4.6999444444444445e-05,
+      "loss": 0.0059,
+      "step": 32420
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.699851851851852e-05,
+      "loss": 0.0044,
+      "step": 32430
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.6997592592592595e-05,
+      "loss": 0.0041,
+      "step": 32440
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.699666666666666e-05,
+      "loss": 0.0054,
+      "step": 32450
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.6995740740740745e-05,
+      "loss": 0.0036,
+      "step": 32460
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.699481481481482e-05,
+      "loss": 0.0051,
+      "step": 32470
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 4.6993888888888895e-05,
+      "loss": 0.0059,
+      "step": 32480
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 4.6992962962962964e-05,
+      "loss": 0.005,
+      "step": 32490
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 4.699203703703704e-05,
+      "loss": 0.0042,
+      "step": 32500
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 4.6991111111111114e-05,
+      "loss": 0.0064,
+      "step": 32510
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 4.699018518518519e-05,
+      "loss": 0.0048,
+      "step": 32520
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 4.6989259259259264e-05,
+      "loss": 0.0053,
+      "step": 32530
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 4.698833333333333e-05,
+      "loss": 0.0059,
+      "step": 32540
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 4.698740740740741e-05,
+      "loss": 0.0073,
+      "step": 32550
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 4.698648148148148e-05,
+      "loss": 0.0046,
+      "step": 32560
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 4.698555555555556e-05,
+      "loss": 0.007,
+      "step": 32570
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 4.698462962962963e-05,
+      "loss": 0.005,
+      "step": 32580
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 4.698370370370371e-05,
+      "loss": 0.0034,
+      "step": 32590
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 4.698277777777778e-05,
+      "loss": 0.0057,
+      "step": 32600
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 4.698185185185186e-05,
+      "loss": 0.0043,
+      "step": 32610
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 4.6980925925925926e-05,
+      "loss": 0.0035,
+      "step": 32620
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 4.698e-05,
+      "loss": 0.0054,
+      "step": 32630
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 4.6979074074074076e-05,
+      "loss": 0.004,
+      "step": 32640
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 4.697814814814815e-05,
+      "loss": 0.0045,
+      "step": 32650
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 4.697722222222222e-05,
+      "loss": 0.0049,
+      "step": 32660
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 4.6976296296296294e-05,
+      "loss": 0.0045,
+      "step": 32670
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 4.6975370370370376e-05,
+      "loss": 0.005,
+      "step": 32680
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 4.697444444444445e-05,
+      "loss": 0.0084,
+      "step": 32690
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 4.697351851851852e-05,
+      "loss": 0.004,
+      "step": 32700
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 4.6972592592592594e-05,
+      "loss": 0.0053,
+      "step": 32710
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 4.697166666666667e-05,
+      "loss": 0.0037,
+      "step": 32720
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 4.6970740740740745e-05,
+      "loss": 0.0054,
+      "step": 32730
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 4.696981481481482e-05,
+      "loss": 0.003,
+      "step": 32740
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 4.696888888888889e-05,
+      "loss": 0.0066,
+      "step": 32750
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 4.696796296296296e-05,
+      "loss": 0.0048,
+      "step": 32760
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 4.696703703703704e-05,
+      "loss": 0.0055,
+      "step": 32770
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 4.696611111111111e-05,
+      "loss": 0.0043,
+      "step": 32780
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 4.696518518518519e-05,
+      "loss": 0.0057,
+      "step": 32790
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 4.696425925925926e-05,
+      "loss": 0.0029,
+      "step": 32800
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 4.696333333333334e-05,
+      "loss": 0.0043,
+      "step": 32810
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 4.696240740740741e-05,
+      "loss": 0.0053,
+      "step": 32820
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 4.696148148148148e-05,
+      "loss": 0.0054,
+      "step": 32830
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 4.696055555555556e-05,
+      "loss": 0.0052,
+      "step": 32840
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 4.695962962962963e-05,
+      "loss": 0.0045,
+      "step": 32850
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 4.695870370370371e-05,
+      "loss": 0.0025,
+      "step": 32860
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 4.6957777777777775e-05,
+      "loss": 0.0043,
+      "step": 32870
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 4.695685185185185e-05,
+      "loss": 0.0046,
+      "step": 32880
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 4.6955925925925925e-05,
+      "loss": 0.0026,
+      "step": 32890
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 4.695500000000001e-05,
+      "loss": 0.0054,
+      "step": 32900
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 4.6954074074074075e-05,
+      "loss": 0.008,
+      "step": 32910
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 4.695314814814815e-05,
+      "loss": 0.0076,
+      "step": 32920
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 4.6952222222222225e-05,
+      "loss": 0.0057,
+      "step": 32930
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 4.69512962962963e-05,
+      "loss": 0.0057,
+      "step": 32940
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 4.6950370370370375e-05,
+      "loss": 0.0068,
+      "step": 32950
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 4.6949444444444444e-05,
+      "loss": 0.0051,
+      "step": 32960
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 4.694851851851852e-05,
+      "loss": 0.0054,
+      "step": 32970
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 4.6947592592592594e-05,
+      "loss": 0.0059,
+      "step": 32980
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 4.694666666666667e-05,
+      "loss": 0.0029,
+      "step": 32990
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 4.6945740740740744e-05,
+      "loss": 0.0034,
+      "step": 33000
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 4.694481481481482e-05,
+      "loss": 0.0045,
+      "step": 33010
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 4.6943888888888894e-05,
+      "loss": 0.0044,
+      "step": 33020
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 4.694296296296297e-05,
+      "loss": 0.005,
+      "step": 33030
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 4.694203703703704e-05,
+      "loss": 0.0051,
+      "step": 33040
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 4.694111111111111e-05,
+      "loss": 0.0075,
+      "step": 33050
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 4.694018518518519e-05,
+      "loss": 0.0028,
+      "step": 33060
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 4.693925925925926e-05,
+      "loss": 0.0045,
+      "step": 33070
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 4.693833333333333e-05,
+      "loss": 0.0047,
+      "step": 33080
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 4.6937407407407406e-05,
+      "loss": 0.0035,
+      "step": 33090
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 4.693648148148148e-05,
+      "loss": 0.0057,
+      "step": 33100
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 4.693555555555556e-05,
+      "loss": 0.0055,
+      "step": 33110
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 4.693462962962963e-05,
+      "loss": 0.0032,
+      "step": 33120
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 4.6933703703703706e-05,
+      "loss": 0.0033,
+      "step": 33130
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 4.693277777777778e-05,
+      "loss": 0.0081,
+      "step": 33140
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 4.6931851851851856e-05,
+      "loss": 0.0041,
+      "step": 33150
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 4.693092592592593e-05,
+      "loss": 0.0081,
+      "step": 33160
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 4.693e-05,
+      "loss": 0.0087,
+      "step": 33170
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 4.6929074074074075e-05,
+      "loss": 0.0056,
+      "step": 33180
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 4.692814814814815e-05,
+      "loss": 0.0041,
+      "step": 33190
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 4.6927222222222225e-05,
+      "loss": 0.0042,
+      "step": 33200
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 4.692629629629629e-05,
+      "loss": 0.0069,
+      "step": 33210
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 4.6925370370370375e-05,
+      "loss": 0.0045,
+      "step": 33220
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 4.692444444444445e-05,
+      "loss": 0.0062,
+      "step": 33230
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 4.6923518518518525e-05,
+      "loss": 0.0056,
+      "step": 33240
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 4.692259259259259e-05,
+      "loss": 0.0058,
+      "step": 33250
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 4.692166666666667e-05,
+      "loss": 0.0092,
+      "step": 33260
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 4.692074074074074e-05,
+      "loss": 0.0045,
+      "step": 33270
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 4.691981481481482e-05,
+      "loss": 0.008,
+      "step": 33280
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 4.691888888888889e-05,
+      "loss": 0.0035,
+      "step": 33290
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 4.691796296296296e-05,
+      "loss": 0.0056,
+      "step": 33300
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 4.691703703703704e-05,
+      "loss": 0.004,
+      "step": 33310
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 4.691611111111111e-05,
+      "loss": 0.0035,
+      "step": 33320
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 4.691518518518519e-05,
+      "loss": 0.0066,
+      "step": 33330
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 4.691425925925926e-05,
+      "loss": 0.0053,
+      "step": 33340
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 4.691333333333334e-05,
+      "loss": 0.0037,
+      "step": 33350
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 4.691240740740741e-05,
+      "loss": 0.0065,
+      "step": 33360
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 4.691148148148149e-05,
+      "loss": 0.0042,
+      "step": 33370
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 4.6910555555555555e-05,
+      "loss": 0.0043,
+      "step": 33380
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 4.690962962962963e-05,
+      "loss": 0.0017,
+      "step": 33390
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 4.6908703703703706e-05,
+      "loss": 0.0069,
+      "step": 33400
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 4.690777777777778e-05,
+      "loss": 0.0036,
+      "step": 33410
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 4.690685185185185e-05,
+      "loss": 0.0049,
+      "step": 33420
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 4.690592592592593e-05,
+      "loss": 0.0075,
+      "step": 33430
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 4.6905000000000006e-05,
+      "loss": 0.0053,
+      "step": 33440
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 4.690407407407408e-05,
+      "loss": 0.0056,
+      "step": 33450
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 4.690314814814815e-05,
+      "loss": 0.0055,
+      "step": 33460
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 4.6902222222222224e-05,
+      "loss": 0.005,
+      "step": 33470
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 4.69012962962963e-05,
+      "loss": 0.0071,
+      "step": 33480
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 4.6900370370370374e-05,
+      "loss": 0.0042,
+      "step": 33490
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 4.689944444444444e-05,
+      "loss": 0.0059,
+      "step": 33500
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 4.689851851851852e-05,
+      "loss": 0.0043,
+      "step": 33510
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 4.689759259259259e-05,
+      "loss": 0.0041,
+      "step": 33520
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 4.689666666666667e-05,
+      "loss": 0.0039,
+      "step": 33530
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 4.689574074074074e-05,
+      "loss": 0.0085,
+      "step": 33540
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 4.689481481481482e-05,
+      "loss": 0.0052,
+      "step": 33550
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 4.689388888888889e-05,
+      "loss": 0.0038,
+      "step": 33560
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 4.689296296296297e-05,
+      "loss": 0.006,
+      "step": 33570
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 4.689203703703704e-05,
+      "loss": 0.0061,
+      "step": 33580
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 4.689111111111111e-05,
+      "loss": 0.0033,
+      "step": 33590
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 4.6890185185185186e-05,
+      "loss": 0.0089,
+      "step": 33600
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 4.688925925925926e-05,
+      "loss": 0.0052,
+      "step": 33610
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 4.6888333333333336e-05,
+      "loss": 0.0063,
+      "step": 33620
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 4.6887407407407405e-05,
+      "loss": 0.0053,
+      "step": 33630
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 4.688648148148148e-05,
+      "loss": 0.0027,
+      "step": 33640
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 4.688555555555556e-05,
+      "loss": 0.005,
+      "step": 33650
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 4.688462962962964e-05,
+      "loss": 0.0041,
+      "step": 33660
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 4.6883703703703705e-05,
+      "loss": 0.0069,
+      "step": 33670
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 4.688277777777778e-05,
+      "loss": 0.0049,
+      "step": 33680
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 4.6881851851851855e-05,
+      "loss": 0.0061,
+      "step": 33690
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 4.688092592592593e-05,
+      "loss": 0.0055,
+      "step": 33700
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 4.688e-05,
+      "loss": 0.0053,
+      "step": 33710
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 4.6879074074074073e-05,
+      "loss": 0.0055,
+      "step": 33720
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 4.687814814814815e-05,
+      "loss": 0.0068,
+      "step": 33730
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 4.6877222222222224e-05,
+      "loss": 0.0078,
+      "step": 33740
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 4.68762962962963e-05,
+      "loss": 0.0033,
+      "step": 33750
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 4.6875370370370374e-05,
+      "loss": 0.0032,
+      "step": 33760
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 4.687444444444445e-05,
+      "loss": 0.0058,
+      "step": 33770
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 4.6873518518518524e-05,
+      "loss": 0.004,
+      "step": 33780
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 4.68725925925926e-05,
+      "loss": 0.0037,
+      "step": 33790
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 4.687166666666667e-05,
+      "loss": 0.003,
+      "step": 33800
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 4.687074074074074e-05,
+      "loss": 0.0049,
+      "step": 33810
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 4.686981481481482e-05,
+      "loss": 0.0046,
+      "step": 33820
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 4.686888888888889e-05,
+      "loss": 0.0048,
+      "step": 33830
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 4.686796296296296e-05,
+      "loss": 0.0062,
+      "step": 33840
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 4.6867037037037036e-05,
+      "loss": 0.0052,
+      "step": 33850
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 4.686611111111111e-05,
+      "loss": 0.0036,
+      "step": 33860
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 4.686518518518519e-05,
+      "loss": 0.0063,
+      "step": 33870
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 4.686425925925926e-05,
+      "loss": 0.0056,
+      "step": 33880
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 4.6863333333333336e-05,
+      "loss": 0.0032,
+      "step": 33890
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 4.686240740740741e-05,
+      "loss": 0.0045,
+      "step": 33900
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 4.6861481481481486e-05,
+      "loss": 0.0053,
+      "step": 33910
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 4.6860555555555554e-05,
+      "loss": 0.0048,
+      "step": 33920
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 4.685962962962963e-05,
+      "loss": 0.0072,
+      "step": 33930
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 4.6858703703703704e-05,
+      "loss": 0.0049,
+      "step": 33940
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 4.685777777777778e-05,
+      "loss": 0.0046,
+      "step": 33950
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 4.6856851851851854e-05,
+      "loss": 0.0028,
+      "step": 33960
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 4.685592592592593e-05,
+      "loss": 0.007,
+      "step": 33970
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 4.6855000000000005e-05,
+      "loss": 0.0039,
+      "step": 33980
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 4.685407407407408e-05,
+      "loss": 0.0052,
+      "step": 33990
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 4.6853148148148155e-05,
+      "loss": 0.0043,
+      "step": 34000
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 4.685222222222222e-05,
+      "loss": 0.005,
+      "step": 34010
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 4.68512962962963e-05,
+      "loss": 0.0071,
+      "step": 34020
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 4.685037037037037e-05,
+      "loss": 0.0072,
+      "step": 34030
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 4.684953703703704e-05,
+      "loss": 0.0079,
+      "step": 34040
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 4.684861111111111e-05,
+      "loss": 0.0034,
+      "step": 34050
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 4.684768518518519e-05,
+      "loss": 0.0069,
+      "step": 34060
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 4.684675925925926e-05,
+      "loss": 0.0037,
+      "step": 34070
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 4.684583333333333e-05,
+      "loss": 0.0052,
+      "step": 34080
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 4.684490740740741e-05,
+      "loss": 0.0046,
+      "step": 34090
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 4.684398148148149e-05,
+      "loss": 0.0048,
+      "step": 34100
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 4.6843055555555557e-05,
+      "loss": 0.0075,
+      "step": 34110
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 4.684212962962963e-05,
+      "loss": 0.0046,
+      "step": 34120
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 4.684120370370371e-05,
+      "loss": 0.0046,
+      "step": 34130
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 4.684027777777778e-05,
+      "loss": 0.0057,
+      "step": 34140
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 4.683935185185186e-05,
+      "loss": 0.0046,
+      "step": 34150
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 4.6838425925925925e-05,
+      "loss": 0.004,
+      "step": 34160
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 4.68375e-05,
+      "loss": 0.0045,
+      "step": 34170
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 4.6836574074074075e-05,
+      "loss": 0.0038,
+      "step": 34180
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 4.683564814814815e-05,
+      "loss": 0.0041,
+      "step": 34190
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 4.6834722222222225e-05,
+      "loss": 0.0045,
+      "step": 34200
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 4.68337962962963e-05,
+      "loss": 0.0041,
+      "step": 34210
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 4.6832870370370375e-05,
+      "loss": 0.0033,
+      "step": 34220
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 4.683194444444445e-05,
+      "loss": 0.0079,
+      "step": 34230
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 4.683101851851852e-05,
+      "loss": 0.0054,
+      "step": 34240
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 4.6830092592592594e-05,
+      "loss": 0.0068,
+      "step": 34250
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 4.682916666666667e-05,
+      "loss": 0.0046,
+      "step": 34260
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 4.6828240740740744e-05,
+      "loss": 0.0048,
+      "step": 34270
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 4.682731481481482e-05,
+      "loss": 0.0043,
+      "step": 34280
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 4.682638888888889e-05,
+      "loss": 0.0049,
+      "step": 34290
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 4.682546296296296e-05,
+      "loss": 0.0039,
+      "step": 34300
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 4.6824537037037044e-05,
+      "loss": 0.0055,
+      "step": 34310
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 4.682361111111111e-05,
+      "loss": 0.0049,
+      "step": 34320
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 4.682268518518519e-05,
+      "loss": 0.0054,
+      "step": 34330
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 4.682175925925926e-05,
+      "loss": 0.0032,
+      "step": 34340
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 4.682083333333334e-05,
+      "loss": 0.0058,
+      "step": 34350
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 4.681990740740741e-05,
+      "loss": 0.0082,
+      "step": 34360
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 4.681898148148148e-05,
+      "loss": 0.0037,
+      "step": 34370
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 4.6818055555555556e-05,
+      "loss": 0.0048,
+      "step": 34380
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 4.681712962962963e-05,
+      "loss": 0.0053,
+      "step": 34390
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 4.6816203703703706e-05,
+      "loss": 0.0029,
+      "step": 34400
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 4.6815277777777774e-05,
+      "loss": 0.0064,
+      "step": 34410
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 4.6814351851851856e-05,
+      "loss": 0.003,
+      "step": 34420
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 4.681342592592593e-05,
+      "loss": 0.0063,
+      "step": 34430
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 4.6812500000000006e-05,
+      "loss": 0.0048,
+      "step": 34440
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 4.6811574074074075e-05,
+      "loss": 0.0071,
+      "step": 34450
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 4.681064814814815e-05,
+      "loss": 0.0048,
+      "step": 34460
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 4.6809722222222225e-05,
+      "loss": 0.0029,
+      "step": 34470
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 4.68087962962963e-05,
+      "loss": 0.0057,
+      "step": 34480
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 4.6807870370370375e-05,
+      "loss": 0.0035,
+      "step": 34490
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 4.680694444444444e-05,
+      "loss": 0.0063,
+      "step": 34500
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 4.680601851851852e-05,
+      "loss": 0.0034,
+      "step": 34510
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 4.68050925925926e-05,
+      "loss": 0.0053,
+      "step": 34520
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 4.680416666666667e-05,
+      "loss": 0.0049,
+      "step": 34530
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 4.680324074074074e-05,
+      "loss": 0.0093,
+      "step": 34540
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 4.680231481481482e-05,
+      "loss": 0.0061,
+      "step": 34550
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 4.680138888888889e-05,
+      "loss": 0.0051,
+      "step": 34560
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 4.680046296296297e-05,
+      "loss": 0.0056,
+      "step": 34570
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 4.679953703703704e-05,
+      "loss": 0.0051,
+      "step": 34580
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 4.679861111111111e-05,
+      "loss": 0.0045,
+      "step": 34590
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 4.679768518518519e-05,
+      "loss": 0.0047,
+      "step": 34600
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 4.679675925925926e-05,
+      "loss": 0.004,
+      "step": 34610
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 4.679583333333333e-05,
+      "loss": 0.004,
+      "step": 34620
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 4.679490740740741e-05,
+      "loss": 0.0102,
+      "step": 34630
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 4.679398148148149e-05,
+      "loss": 0.0049,
+      "step": 34640
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 4.679305555555556e-05,
+      "loss": 0.0041,
+      "step": 34650
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 4.679212962962963e-05,
+      "loss": 0.0067,
+      "step": 34660
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 4.6791203703703705e-05,
+      "loss": 0.0047,
+      "step": 34670
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 4.679027777777778e-05,
+      "loss": 0.006,
+      "step": 34680
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 4.6789351851851856e-05,
+      "loss": 0.0038,
+      "step": 34690
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 4.678842592592593e-05,
+      "loss": 0.0083,
+      "step": 34700
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 4.67875e-05,
+      "loss": 0.0042,
+      "step": 34710
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 4.6786574074074074e-05,
+      "loss": 0.0051,
+      "step": 34720
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 4.678564814814815e-05,
+      "loss": 0.0063,
+      "step": 34730
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 4.6784722222222224e-05,
+      "loss": 0.0036,
+      "step": 34740
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 4.67837962962963e-05,
+      "loss": 0.0069,
+      "step": 34750
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 4.6782870370370374e-05,
+      "loss": 0.0068,
+      "step": 34760
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 4.678194444444445e-05,
+      "loss": 0.0059,
+      "step": 34770
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 4.6781018518518524e-05,
+      "loss": 0.0056,
+      "step": 34780
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 4.678009259259259e-05,
+      "loss": 0.0041,
+      "step": 34790
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 4.677916666666667e-05,
+      "loss": 0.0091,
+      "step": 34800
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 4.677824074074074e-05,
+      "loss": 0.0036,
+      "step": 34810
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 4.677731481481482e-05,
+      "loss": 0.0047,
+      "step": 34820
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 4.6776388888888886e-05,
+      "loss": 0.0033,
+      "step": 34830
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 4.677546296296296e-05,
+      "loss": 0.0055,
+      "step": 34840
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 4.677453703703704e-05,
+      "loss": 0.0056,
+      "step": 34850
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 4.677361111111112e-05,
+      "loss": 0.0033,
+      "step": 34860
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 4.6772685185185186e-05,
+      "loss": 0.0042,
+      "step": 34870
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 4.677175925925926e-05,
+      "loss": 0.0041,
+      "step": 34880
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 4.6770833333333336e-05,
+      "loss": 0.0021,
+      "step": 34890
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 4.676990740740741e-05,
+      "loss": 0.0088,
+      "step": 34900
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 4.6768981481481486e-05,
+      "loss": 0.0038,
+      "step": 34910
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 4.6768055555555555e-05,
+      "loss": 0.0052,
+      "step": 34920
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 4.676712962962963e-05,
+      "loss": 0.0039,
+      "step": 34930
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 4.6766203703703705e-05,
+      "loss": 0.0044,
+      "step": 34940
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 4.676527777777778e-05,
+      "loss": 0.0051,
+      "step": 34950
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 4.6764351851851855e-05,
+      "loss": 0.0041,
+      "step": 34960
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 4.676342592592593e-05,
+      "loss": 0.0056,
+      "step": 34970
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 4.6762500000000005e-05,
+      "loss": 0.004,
+      "step": 34980
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 4.676157407407408e-05,
+      "loss": 0.0048,
+      "step": 34990
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 4.676064814814815e-05,
+      "loss": 0.0049,
+      "step": 35000
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 4.6759722222222223e-05,
+      "loss": 0.0035,
+      "step": 35010
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 4.67587962962963e-05,
+      "loss": 0.0045,
+      "step": 35020
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 4.6757870370370374e-05,
+      "loss": 0.0067,
+      "step": 35030
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 4.675694444444444e-05,
+      "loss": 0.0034,
+      "step": 35040
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 4.675601851851852e-05,
+      "loss": 0.0046,
+      "step": 35050
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 4.67550925925926e-05,
+      "loss": 0.0051,
+      "step": 35060
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 4.6754166666666674e-05,
+      "loss": 0.0067,
+      "step": 35070
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 4.675324074074074e-05,
+      "loss": 0.0065,
+      "step": 35080
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 4.675231481481482e-05,
+      "loss": 0.0078,
+      "step": 35090
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 4.675138888888889e-05,
+      "loss": 0.0061,
+      "step": 35100
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 4.675046296296297e-05,
+      "loss": 0.0024,
+      "step": 35110
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 4.674953703703704e-05,
+      "loss": 0.0036,
+      "step": 35120
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 4.674861111111111e-05,
+      "loss": 0.0078,
+      "step": 35130
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 4.6747685185185186e-05,
+      "loss": 0.0056,
+      "step": 35140
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 4.674675925925926e-05,
+      "loss": 0.0066,
+      "step": 35150
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 4.6745833333333336e-05,
+      "loss": 0.0058,
+      "step": 35160
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 4.674490740740741e-05,
+      "loss": 0.0036,
+      "step": 35170
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 4.6743981481481486e-05,
+      "loss": 0.0054,
+      "step": 35180
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 4.674305555555556e-05,
+      "loss": 0.0032,
+      "step": 35190
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 4.6742129629629636e-05,
+      "loss": 0.0051,
+      "step": 35200
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 4.6741203703703704e-05,
+      "loss": 0.0051,
+      "step": 35210
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 4.674027777777778e-05,
+      "loss": 0.0053,
+      "step": 35220
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 4.6739351851851854e-05,
+      "loss": 0.0065,
+      "step": 35230
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 4.673842592592593e-05,
+      "loss": 0.0032,
+      "step": 35240
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 4.67375e-05,
+      "loss": 0.0047,
+      "step": 35250
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 4.673657407407407e-05,
+      "loss": 0.0042,
+      "step": 35260
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 4.673564814814815e-05,
+      "loss": 0.0062,
+      "step": 35270
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 4.673472222222223e-05,
+      "loss": 0.0077,
+      "step": 35280
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 4.67337962962963e-05,
+      "loss": 0.0023,
+      "step": 35290
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 4.673287037037037e-05,
+      "loss": 0.0054,
+      "step": 35300
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 4.673194444444445e-05,
+      "loss": 0.0046,
+      "step": 35310
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 4.673101851851852e-05,
+      "loss": 0.0056,
+      "step": 35320
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 4.67300925925926e-05,
+      "loss": 0.0045,
+      "step": 35330
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 4.6729166666666666e-05,
+      "loss": 0.0054,
+      "step": 35340
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 4.672824074074074e-05,
+      "loss": 0.0073,
+      "step": 35350
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 4.6727314814814816e-05,
+      "loss": 0.0047,
+      "step": 35360
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 4.672638888888889e-05,
+      "loss": 0.0074,
+      "step": 35370
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 4.672546296296296e-05,
+      "loss": 0.0046,
+      "step": 35380
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 4.672453703703704e-05,
+      "loss": 0.0039,
+      "step": 35390
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 4.672361111111112e-05,
+      "loss": 0.0058,
+      "step": 35400
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 4.672268518518519e-05,
+      "loss": 0.0021,
+      "step": 35410
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 4.672175925925926e-05,
+      "loss": 0.0041,
+      "step": 35420
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 4.6720833333333335e-05,
+      "loss": 0.0066,
+      "step": 35430
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 4.671990740740741e-05,
+      "loss": 0.0057,
+      "step": 35440
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 4.6718981481481485e-05,
+      "loss": 0.0045,
+      "step": 35450
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 4.6718055555555553e-05,
+      "loss": 0.006,
+      "step": 35460
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 4.671712962962963e-05,
+      "loss": 0.0039,
+      "step": 35470
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 4.6716203703703704e-05,
+      "loss": 0.0052,
+      "step": 35480
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 4.671527777777778e-05,
+      "loss": 0.006,
+      "step": 35490
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 4.6714351851851854e-05,
+      "loss": 0.0048,
+      "step": 35500
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 4.671342592592593e-05,
+      "loss": 0.0056,
+      "step": 35510
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 4.6712500000000004e-05,
+      "loss": 0.0038,
+      "step": 35520
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 4.671157407407408e-05,
+      "loss": 0.0038,
+      "step": 35530
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 4.6710648148148154e-05,
+      "loss": 0.0072,
+      "step": 35540
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 4.670972222222222e-05,
+      "loss": 0.0046,
+      "step": 35550
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 4.67087962962963e-05,
+      "loss": 0.0051,
+      "step": 35560
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 4.670787037037037e-05,
+      "loss": 0.006,
+      "step": 35570
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 4.670694444444445e-05,
+      "loss": 0.0051,
+      "step": 35580
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 4.6706018518518516e-05,
+      "loss": 0.0042,
+      "step": 35590
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 4.67050925925926e-05,
+      "loss": 0.0057,
+      "step": 35600
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 4.670416666666667e-05,
+      "loss": 0.0039,
+      "step": 35610
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 4.670324074074075e-05,
+      "loss": 0.0061,
+      "step": 35620
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 4.6702314814814816e-05,
+      "loss": 0.0051,
+      "step": 35630
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 4.670138888888889e-05,
+      "loss": 0.0056,
+      "step": 35640
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 4.6700462962962966e-05,
+      "loss": 0.0075,
+      "step": 35650
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 4.669953703703704e-05,
+      "loss": 0.0057,
+      "step": 35660
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 4.669861111111111e-05,
+      "loss": 0.0041,
+      "step": 35670
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 4.6697685185185184e-05,
+      "loss": 0.0052,
+      "step": 35680
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 4.669675925925926e-05,
+      "loss": 0.0037,
+      "step": 35690
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 4.6695833333333334e-05,
+      "loss": 0.0065,
+      "step": 35700
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 4.669490740740741e-05,
+      "loss": 0.0062,
+      "step": 35710
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 4.6693981481481485e-05,
+      "loss": 0.0059,
+      "step": 35720
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 4.669305555555556e-05,
+      "loss": 0.0038,
+      "step": 35730
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 4.6692129629629635e-05,
+      "loss": 0.0068,
+      "step": 35740
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 4.669120370370371e-05,
+      "loss": 0.0036,
+      "step": 35750
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 4.669027777777778e-05,
+      "loss": 0.0036,
+      "step": 35760
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 4.668935185185185e-05,
+      "loss": 0.0065,
+      "step": 35770
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 4.668842592592593e-05,
+      "loss": 0.0056,
+      "step": 35780
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 4.66875e-05,
+      "loss": 0.006,
+      "step": 35790
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 4.668657407407407e-05,
+      "loss": 0.0044,
+      "step": 35800
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 4.6685648148148147e-05,
+      "loss": 0.0052,
+      "step": 35810
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 4.668472222222223e-05,
+      "loss": 0.0065,
+      "step": 35820
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 4.6683796296296303e-05,
+      "loss": 0.0055,
+      "step": 35830
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 4.668287037037037e-05,
+      "loss": 0.0045,
+      "step": 35840
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 4.668194444444445e-05,
+      "loss": 0.0069,
+      "step": 35850
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 4.668101851851852e-05,
+      "loss": 0.0026,
+      "step": 35860
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 4.66800925925926e-05,
+      "loss": 0.0049,
+      "step": 35870
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 4.6679166666666665e-05,
+      "loss": 0.0061,
+      "step": 35880
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 4.667824074074074e-05,
+      "loss": 0.0044,
+      "step": 35890
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 4.6677314814814815e-05,
+      "loss": 0.0033,
+      "step": 35900
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 4.667638888888889e-05,
+      "loss": 0.0044,
+      "step": 35910
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 4.6675462962962965e-05,
+      "loss": 0.005,
+      "step": 35920
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 4.667453703703704e-05,
+      "loss": 0.0043,
+      "step": 35930
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 4.6673611111111115e-05,
+      "loss": 0.0061,
+      "step": 35940
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 4.667268518518519e-05,
+      "loss": 0.0053,
+      "step": 35950
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 4.6671759259259266e-05,
+      "loss": 0.0037,
+      "step": 35960
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 4.6670833333333334e-05,
+      "loss": 0.0059,
+      "step": 35970
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 4.666990740740741e-05,
+      "loss": 0.0054,
+      "step": 35980
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 4.6668981481481484e-05,
+      "loss": 0.003,
+      "step": 35990
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 4.666805555555556e-05,
+      "loss": 0.0032,
+      "step": 36000
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 4.666712962962963e-05,
+      "loss": 0.0072,
+      "step": 36010
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 4.66662037037037e-05,
+      "loss": 0.003,
+      "step": 36020
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 4.6665277777777784e-05,
+      "loss": 0.0071,
+      "step": 36030
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 4.666435185185186e-05,
+      "loss": 0.0049,
+      "step": 36040
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 4.666342592592593e-05,
+      "loss": 0.0048,
+      "step": 36050
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 4.66625e-05,
+      "loss": 0.0041,
+      "step": 36060
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 4.666157407407408e-05,
+      "loss": 0.0063,
+      "step": 36070
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 4.666064814814815e-05,
+      "loss": 0.0038,
+      "step": 36080
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 4.665972222222222e-05,
+      "loss": 0.0052,
+      "step": 36090
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 4.6658796296296296e-05,
+      "loss": 0.0037,
+      "step": 36100
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 4.665787037037037e-05,
+      "loss": 0.0045,
+      "step": 36110
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 4.6656944444444446e-05,
+      "loss": 0.0032,
+      "step": 36120
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 4.665601851851852e-05,
+      "loss": 0.0042,
+      "step": 36130
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 4.6655092592592596e-05,
+      "loss": 0.005,
+      "step": 36140
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 4.665425925925926e-05,
+      "loss": 0.0065,
+      "step": 36150
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 4.6653333333333336e-05,
+      "loss": 0.0043,
+      "step": 36160
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 4.665240740740741e-05,
+      "loss": 0.005,
+      "step": 36170
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 4.6651481481481486e-05,
+      "loss": 0.0076,
+      "step": 36180
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 4.665055555555556e-05,
+      "loss": 0.003,
+      "step": 36190
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 4.664962962962963e-05,
+      "loss": 0.0038,
+      "step": 36200
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 4.6648703703703705e-05,
+      "loss": 0.0031,
+      "step": 36210
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 4.664777777777778e-05,
+      "loss": 0.0047,
+      "step": 36220
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 4.6646851851851855e-05,
+      "loss": 0.006,
+      "step": 36230
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 4.664592592592592e-05,
+      "loss": 0.0043,
+      "step": 36240
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 4.6645e-05,
+      "loss": 0.0034,
+      "step": 36250
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 4.664407407407408e-05,
+      "loss": 0.0068,
+      "step": 36260
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 4.6643148148148155e-05,
+      "loss": 0.0042,
+      "step": 36270
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 4.664222222222222e-05,
+      "loss": 0.0044,
+      "step": 36280
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 4.66412962962963e-05,
+      "loss": 0.0056,
+      "step": 36290
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 4.664037037037037e-05,
+      "loss": 0.0065,
+      "step": 36300
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 4.663944444444445e-05,
+      "loss": 0.0065,
+      "step": 36310
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 4.6638518518518523e-05,
+      "loss": 0.0048,
+      "step": 36320
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 4.663759259259259e-05,
+      "loss": 0.0059,
+      "step": 36330
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 4.663666666666667e-05,
+      "loss": 0.0055,
+      "step": 36340
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 4.663574074074074e-05,
+      "loss": 0.006,
+      "step": 36350
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 4.663481481481482e-05,
+      "loss": 0.0074,
+      "step": 36360
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 4.663388888888889e-05,
+      "loss": 0.0033,
+      "step": 36370
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 4.663296296296297e-05,
+      "loss": 0.0036,
+      "step": 36380
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 4.663203703703704e-05,
+      "loss": 0.008,
+      "step": 36390
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 4.663111111111112e-05,
+      "loss": 0.004,
+      "step": 36400
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 4.6630185185185185e-05,
+      "loss": 0.0039,
+      "step": 36410
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 4.662925925925926e-05,
+      "loss": 0.0055,
+      "step": 36420
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 4.6628333333333336e-05,
+      "loss": 0.0046,
+      "step": 36430
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 4.662740740740741e-05,
+      "loss": 0.005,
+      "step": 36440
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 4.662648148148148e-05,
+      "loss": 0.0065,
+      "step": 36450
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 4.6625555555555554e-05,
+      "loss": 0.0057,
+      "step": 36460
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 4.662462962962963e-05,
+      "loss": 0.006,
+      "step": 36470
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 4.662370370370371e-05,
+      "loss": 0.0046,
+      "step": 36480
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 4.662277777777778e-05,
+      "loss": 0.0034,
+      "step": 36490
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 4.6621851851851854e-05,
+      "loss": 0.0055,
+      "step": 36500
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 4.662092592592593e-05,
+      "loss": 0.0061,
+      "step": 36510
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 4.6620000000000004e-05,
+      "loss": 0.0052,
+      "step": 36520
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 4.661907407407408e-05,
+      "loss": 0.0038,
+      "step": 36530
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 4.661814814814815e-05,
+      "loss": 0.0047,
+      "step": 36540
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 4.661722222222222e-05,
+      "loss": 0.0058,
+      "step": 36550
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 4.66162962962963e-05,
+      "loss": 0.0058,
+      "step": 36560
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 4.661537037037037e-05,
+      "loss": 0.0039,
+      "step": 36570
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 4.661444444444444e-05,
+      "loss": 0.0063,
+      "step": 36580
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 4.661351851851852e-05,
+      "loss": 0.0066,
+      "step": 36590
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 4.66125925925926e-05,
+      "loss": 0.0052,
+      "step": 36600
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 4.661166666666667e-05,
+      "loss": 0.0043,
+      "step": 36610
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 4.661074074074074e-05,
+      "loss": 0.0045,
+      "step": 36620
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 4.6609814814814816e-05,
+      "loss": 0.0041,
+      "step": 36630
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 4.660888888888889e-05,
+      "loss": 0.0066,
+      "step": 36640
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 4.6607962962962966e-05,
+      "loss": 0.0024,
+      "step": 36650
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 4.6607037037037035e-05,
+      "loss": 0.0055,
+      "step": 36660
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 4.660611111111111e-05,
+      "loss": 0.0038,
+      "step": 36670
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 4.6605185185185185e-05,
+      "loss": 0.0084,
+      "step": 36680
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 4.660425925925927e-05,
+      "loss": 0.0052,
+      "step": 36690
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 4.6603333333333335e-05,
+      "loss": 0.0048,
+      "step": 36700
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 4.660240740740741e-05,
+      "loss": 0.0055,
+      "step": 36710
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 4.6601481481481485e-05,
+      "loss": 0.0043,
+      "step": 36720
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 4.660055555555556e-05,
+      "loss": 0.0035,
+      "step": 36730
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 4.6599629629629635e-05,
+      "loss": 0.0073,
+      "step": 36740
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 4.6598703703703703e-05,
+      "loss": 0.0055,
+      "step": 36750
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 4.659777777777778e-05,
+      "loss": 0.0037,
+      "step": 36760
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 4.6596851851851854e-05,
+      "loss": 0.0055,
+      "step": 36770
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 4.659592592592593e-05,
+      "loss": 0.0068,
+      "step": 36780
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 4.6595e-05,
+      "loss": 0.0057,
+      "step": 36790
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 4.659407407407408e-05,
+      "loss": 0.0053,
+      "step": 36800
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 4.6593148148148154e-05,
+      "loss": 0.0043,
+      "step": 36810
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 4.659222222222223e-05,
+      "loss": 0.0044,
+      "step": 36820
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 4.65912962962963e-05,
+      "loss": 0.0094,
+      "step": 36830
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 4.659037037037037e-05,
+      "loss": 0.0053,
+      "step": 36840
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 4.658944444444445e-05,
+      "loss": 0.0072,
+      "step": 36850
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 4.658851851851852e-05,
+      "loss": 0.0039,
+      "step": 36860
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 4.658759259259259e-05,
+      "loss": 0.0055,
+      "step": 36870
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 4.6586666666666666e-05,
+      "loss": 0.0051,
+      "step": 36880
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 4.658574074074074e-05,
+      "loss": 0.0038,
+      "step": 36890
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 4.6584814814814816e-05,
+      "loss": 0.0068,
+      "step": 36900
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 4.658388888888889e-05,
+      "loss": 0.0044,
+      "step": 36910
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 4.6582962962962966e-05,
+      "loss": 0.004,
+      "step": 36920
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 4.658203703703704e-05,
+      "loss": 0.0056,
+      "step": 36930
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 4.6581111111111116e-05,
+      "loss": 0.0054,
+      "step": 36940
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 4.658018518518519e-05,
+      "loss": 0.0046,
+      "step": 36950
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 4.657925925925926e-05,
+      "loss": 0.0047,
+      "step": 36960
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 4.6578333333333334e-05,
+      "loss": 0.0079,
+      "step": 36970
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 4.657740740740741e-05,
+      "loss": 0.0061,
+      "step": 36980
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 4.6576481481481484e-05,
+      "loss": 0.0037,
+      "step": 36990
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 4.657555555555555e-05,
+      "loss": 0.0065,
+      "step": 37000
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 4.657462962962963e-05,
+      "loss": 0.0038,
+      "step": 37010
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 4.657370370370371e-05,
+      "loss": 0.0052,
+      "step": 37020
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 4.6572777777777785e-05,
+      "loss": 0.0056,
+      "step": 37030
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 4.657185185185185e-05,
+      "loss": 0.0084,
+      "step": 37040
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 4.657092592592593e-05,
+      "loss": 0.0058,
+      "step": 37050
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 4.657e-05,
+      "loss": 0.0037,
+      "step": 37060
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 4.656907407407408e-05,
+      "loss": 0.004,
+      "step": 37070
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 4.6568148148148146e-05,
+      "loss": 0.0036,
+      "step": 37080
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 4.656722222222222e-05,
+      "loss": 0.0039,
+      "step": 37090
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 4.6566296296296296e-05,
+      "loss": 0.0073,
+      "step": 37100
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 4.656537037037037e-05,
+      "loss": 0.0044,
+      "step": 37110
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 4.6564444444444447e-05,
+      "loss": 0.008,
+      "step": 37120
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 4.656351851851852e-05,
+      "loss": 0.004,
+      "step": 37130
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 4.65625925925926e-05,
+      "loss": 0.0062,
+      "step": 37140
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 4.656166666666667e-05,
+      "loss": 0.0039,
+      "step": 37150
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 4.656074074074075e-05,
+      "loss": 0.0051,
+      "step": 37160
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 4.6559814814814815e-05,
+      "loss": 0.0037,
+      "step": 37170
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 4.655888888888889e-05,
+      "loss": 0.0051,
+      "step": 37180
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 4.6557962962962965e-05,
+      "loss": 0.0059,
+      "step": 37190
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 4.655703703703704e-05,
+      "loss": 0.0035,
+      "step": 37200
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 4.655611111111111e-05,
+      "loss": 0.0029,
+      "step": 37210
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 4.6555185185185184e-05,
+      "loss": 0.0037,
+      "step": 37220
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 4.6554259259259265e-05,
+      "loss": 0.0032,
+      "step": 37230
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 4.655333333333334e-05,
+      "loss": 0.0038,
+      "step": 37240
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 4.655240740740741e-05,
+      "loss": 0.006,
+      "step": 37250
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 4.6551481481481484e-05,
+      "loss": 0.003,
+      "step": 37260
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 4.655055555555556e-05,
+      "loss": 0.003,
+      "step": 37270
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 4.6549629629629634e-05,
+      "loss": 0.0073,
+      "step": 37280
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 4.65487037037037e-05,
+      "loss": 0.0033,
+      "step": 37290
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 4.654777777777778e-05,
+      "loss": 0.0052,
+      "step": 37300
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 4.654685185185185e-05,
+      "loss": 0.006,
+      "step": 37310
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 4.654592592592593e-05,
+      "loss": 0.0072,
+      "step": 37320
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 4.6545e-05,
+      "loss": 0.0054,
+      "step": 37330
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 4.654407407407408e-05,
+      "loss": 0.0052,
+      "step": 37340
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 4.654314814814815e-05,
+      "loss": 0.0028,
+      "step": 37350
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 4.654222222222223e-05,
+      "loss": 0.0046,
+      "step": 37360
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 4.65412962962963e-05,
+      "loss": 0.0046,
+      "step": 37370
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 4.654037037037037e-05,
+      "loss": 0.0073,
+      "step": 37380
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 4.6539444444444446e-05,
+      "loss": 0.0067,
+      "step": 37390
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 4.653851851851852e-05,
+      "loss": 0.0094,
+      "step": 37400
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 4.6537592592592596e-05,
+      "loss": 0.0023,
+      "step": 37410
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 4.6536666666666664e-05,
+      "loss": 0.0038,
+      "step": 37420
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 4.653574074074074e-05,
+      "loss": 0.0065,
+      "step": 37430
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 4.6534814814814814e-05,
+      "loss": 0.0041,
+      "step": 37440
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 4.6533888888888896e-05,
+      "loss": 0.0044,
+      "step": 37450
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 4.6532962962962965e-05,
+      "loss": 0.0049,
+      "step": 37460
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 4.653203703703704e-05,
+      "loss": 0.0048,
+      "step": 37470
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 4.6531111111111115e-05,
+      "loss": 0.0051,
+      "step": 37480
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 4.653018518518519e-05,
+      "loss": 0.0043,
+      "step": 37490
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 4.652925925925926e-05,
+      "loss": 0.0059,
+      "step": 37500
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 4.652833333333333e-05,
+      "loss": 0.0061,
+      "step": 37510
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 4.652740740740741e-05,
+      "loss": 0.0064,
+      "step": 37520
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 4.652648148148148e-05,
+      "loss": 0.0044,
+      "step": 37530
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 4.652555555555556e-05,
+      "loss": 0.0056,
+      "step": 37540
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 4.6524629629629627e-05,
+      "loss": 0.0035,
+      "step": 37550
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 4.652370370370371e-05,
+      "loss": 0.0039,
+      "step": 37560
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 4.6522777777777783e-05,
+      "loss": 0.005,
+      "step": 37570
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 4.652185185185186e-05,
+      "loss": 0.0039,
+      "step": 37580
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 4.652092592592593e-05,
+      "loss": 0.0048,
+      "step": 37590
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 4.652e-05,
+      "loss": 0.0045,
+      "step": 37600
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 4.651907407407408e-05,
+      "loss": 0.0053,
+      "step": 37610
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 4.651814814814815e-05,
+      "loss": 0.0054,
+      "step": 37620
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 4.651722222222222e-05,
+      "loss": 0.0059,
+      "step": 37630
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 4.6516296296296295e-05,
+      "loss": 0.0075,
+      "step": 37640
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 4.651537037037037e-05,
+      "loss": 0.0061,
+      "step": 37650
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 4.651444444444445e-05,
+      "loss": 0.0045,
+      "step": 37660
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 4.651351851851852e-05,
+      "loss": 0.0061,
+      "step": 37670
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 4.6512592592592596e-05,
+      "loss": 0.0057,
+      "step": 37680
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 4.651166666666667e-05,
+      "loss": 0.0036,
+      "step": 37690
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 4.6510740740740746e-05,
+      "loss": 0.0053,
+      "step": 37700
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 4.6509814814814814e-05,
+      "loss": 0.0037,
+      "step": 37710
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 4.650888888888889e-05,
+      "loss": 0.0046,
+      "step": 37720
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 4.6507962962962964e-05,
+      "loss": 0.0056,
+      "step": 37730
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 4.650703703703704e-05,
+      "loss": 0.0051,
+      "step": 37740
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 4.6506111111111114e-05,
+      "loss": 0.0063,
+      "step": 37750
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 4.650518518518518e-05,
+      "loss": 0.0062,
+      "step": 37760
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 4.6504259259259264e-05,
+      "loss": 0.0043,
+      "step": 37770
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 4.650333333333334e-05,
+      "loss": 0.0056,
+      "step": 37780
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 4.6502407407407414e-05,
+      "loss": 0.0067,
+      "step": 37790
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 4.650148148148148e-05,
+      "loss": 0.0034,
+      "step": 37800
+    },
+    {
+      "epoch": 7.0,
+      "eval_loss": 0.012877647764980793,
+      "eval_runtime": 153.3867,
+      "eval_samples_per_second": 26.013,
+      "eval_steps_per_second": 3.253,
+      "step": 37800
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 4.650055555555556e-05,
+      "loss": 0.0037,
+      "step": 37810
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 4.649962962962963e-05,
+      "loss": 0.0059,
+      "step": 37820
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 4.649870370370371e-05,
+      "loss": 0.0031,
+      "step": 37830
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 4.6497777777777776e-05,
+      "loss": 0.0047,
+      "step": 37840
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 4.649685185185185e-05,
+      "loss": 0.003,
+      "step": 37850
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 4.6495925925925926e-05,
+      "loss": 0.0064,
+      "step": 37860
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 4.6495e-05,
+      "loss": 0.0042,
+      "step": 37870
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 4.6494074074074076e-05,
+      "loss": 0.0019,
+      "step": 37880
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 4.649314814814815e-05,
+      "loss": 0.0054,
+      "step": 37890
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 4.6492222222222226e-05,
+      "loss": 0.0059,
+      "step": 37900
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 4.64912962962963e-05,
+      "loss": 0.0043,
+      "step": 37910
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 4.649037037037037e-05,
+      "loss": 0.0045,
+      "step": 37920
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 4.6489444444444445e-05,
+      "loss": 0.004,
+      "step": 37930
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 4.648851851851852e-05,
+      "loss": 0.0035,
+      "step": 37940
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 4.6487592592592595e-05,
+      "loss": 0.0038,
+      "step": 37950
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 4.648666666666667e-05,
+      "loss": 0.0034,
+      "step": 37960
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 4.648574074074074e-05,
+      "loss": 0.0053,
+      "step": 37970
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 4.648481481481481e-05,
+      "loss": 0.0043,
+      "step": 37980
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 4.6483888888888895e-05,
+      "loss": 0.0035,
+      "step": 37990
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 4.648296296296297e-05,
+      "loss": 0.0031,
+      "step": 38000
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 4.648203703703704e-05,
+      "loss": 0.0035,
+      "step": 38010
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 4.6481111111111114e-05,
+      "loss": 0.0066,
+      "step": 38020
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 4.648018518518519e-05,
+      "loss": 0.0024,
+      "step": 38030
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 4.6479259259259264e-05,
+      "loss": 0.003,
+      "step": 38040
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 4.647833333333333e-05,
+      "loss": 0.0044,
+      "step": 38050
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 4.647740740740741e-05,
+      "loss": 0.0048,
+      "step": 38060
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 4.647648148148148e-05,
+      "loss": 0.0046,
+      "step": 38070
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 4.647555555555556e-05,
+      "loss": 0.003,
+      "step": 38080
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 4.647462962962963e-05,
+      "loss": 0.0051,
+      "step": 38090
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 4.647370370370371e-05,
+      "loss": 0.0027,
+      "step": 38100
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 4.647277777777778e-05,
+      "loss": 0.0038,
+      "step": 38110
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 4.647185185185186e-05,
+      "loss": 0.0049,
+      "step": 38120
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 4.6470925925925926e-05,
+      "loss": 0.0058,
+      "step": 38130
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 4.647e-05,
+      "loss": 0.0077,
+      "step": 38140
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 4.6469074074074076e-05,
+      "loss": 0.0036,
+      "step": 38150
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 4.646814814814815e-05,
+      "loss": 0.0045,
+      "step": 38160
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 4.6467222222222226e-05,
+      "loss": 0.0057,
+      "step": 38170
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 4.6466296296296294e-05,
+      "loss": 0.0061,
+      "step": 38180
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 4.646537037037037e-05,
+      "loss": 0.0052,
+      "step": 38190
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 4.646444444444445e-05,
+      "loss": 0.0049,
+      "step": 38200
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 4.6463518518518526e-05,
+      "loss": 0.0042,
+      "step": 38210
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 4.6462592592592594e-05,
+      "loss": 0.0062,
+      "step": 38220
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 4.646166666666667e-05,
+      "loss": 0.0062,
+      "step": 38230
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 4.6460740740740744e-05,
+      "loss": 0.0042,
+      "step": 38240
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 4.645981481481482e-05,
+      "loss": 0.0045,
+      "step": 38250
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 4.645888888888889e-05,
+      "loss": 0.0037,
+      "step": 38260
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 4.645796296296296e-05,
+      "loss": 0.0031,
+      "step": 38270
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 4.645703703703704e-05,
+      "loss": 0.0042,
+      "step": 38280
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 4.645611111111111e-05,
+      "loss": 0.0042,
+      "step": 38290
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 4.645518518518519e-05,
+      "loss": 0.0047,
+      "step": 38300
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 4.645425925925926e-05,
+      "loss": 0.0051,
+      "step": 38310
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 4.645333333333334e-05,
+      "loss": 0.0041,
+      "step": 38320
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 4.645240740740741e-05,
+      "loss": 0.0031,
+      "step": 38330
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 4.645148148148148e-05,
+      "loss": 0.0053,
+      "step": 38340
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 4.6450555555555556e-05,
+      "loss": 0.0043,
+      "step": 38350
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 4.644962962962963e-05,
+      "loss": 0.0039,
+      "step": 38360
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 4.6448703703703707e-05,
+      "loss": 0.0033,
+      "step": 38370
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 4.644777777777778e-05,
+      "loss": 0.0033,
+      "step": 38380
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 4.644685185185185e-05,
+      "loss": 0.0054,
+      "step": 38390
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 4.6445925925925925e-05,
+      "loss": 0.0061,
+      "step": 38400
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 4.6445e-05,
+      "loss": 0.0064,
+      "step": 38410
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 4.644407407407408e-05,
+      "loss": 0.0051,
+      "step": 38420
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 4.644314814814815e-05,
+      "loss": 0.0024,
+      "step": 38430
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 4.6442222222222225e-05,
+      "loss": 0.0036,
+      "step": 38440
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 4.64412962962963e-05,
+      "loss": 0.0042,
+      "step": 38450
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 4.6440370370370375e-05,
+      "loss": 0.0023,
+      "step": 38460
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 4.6439444444444444e-05,
+      "loss": 0.0074,
+      "step": 38470
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 4.643851851851852e-05,
+      "loss": 0.007,
+      "step": 38480
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 4.6437592592592594e-05,
+      "loss": 0.0052,
+      "step": 38490
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 4.643666666666667e-05,
+      "loss": 0.0031,
+      "step": 38500
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 4.6435740740740744e-05,
+      "loss": 0.0036,
+      "step": 38510
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 4.643481481481481e-05,
+      "loss": 0.003,
+      "step": 38520
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 4.6433981481481484e-05,
+      "loss": 0.0047,
+      "step": 38530
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 4.643305555555556e-05,
+      "loss": 0.0057,
+      "step": 38540
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 4.6432129629629634e-05,
+      "loss": 0.0059,
+      "step": 38550
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 4.643120370370371e-05,
+      "loss": 0.0049,
+      "step": 38560
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 4.6430277777777784e-05,
+      "loss": 0.0054,
+      "step": 38570
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 4.642935185185185e-05,
+      "loss": 0.0059,
+      "step": 38580
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 4.642842592592593e-05,
+      "loss": 0.0043,
+      "step": 38590
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 4.64275e-05,
+      "loss": 0.004,
+      "step": 38600
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 4.642657407407408e-05,
+      "loss": 0.0039,
+      "step": 38610
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 4.6425648148148146e-05,
+      "loss": 0.0021,
+      "step": 38620
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 4.642472222222222e-05,
+      "loss": 0.005,
+      "step": 38630
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 4.6423796296296296e-05,
+      "loss": 0.0031,
+      "step": 38640
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 4.642287037037038e-05,
+      "loss": 0.0049,
+      "step": 38650
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 4.6421944444444446e-05,
+      "loss": 0.0033,
+      "step": 38660
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 4.642101851851852e-05,
+      "loss": 0.0029,
+      "step": 38670
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 4.6420092592592596e-05,
+      "loss": 0.0034,
+      "step": 38680
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 4.641916666666667e-05,
+      "loss": 0.0036,
+      "step": 38690
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 4.6418240740740746e-05,
+      "loss": 0.0072,
+      "step": 38700
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 4.6417314814814814e-05,
+      "loss": 0.0031,
+      "step": 38710
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 4.641638888888889e-05,
+      "loss": 0.0036,
+      "step": 38720
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 4.6415462962962964e-05,
+      "loss": 0.0045,
+      "step": 38730
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 4.641453703703704e-05,
+      "loss": 0.0055,
+      "step": 38740
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 4.6413611111111115e-05,
+      "loss": 0.0053,
+      "step": 38750
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 4.641268518518519e-05,
+      "loss": 0.0058,
+      "step": 38760
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 4.6411759259259265e-05,
+      "loss": 0.0036,
+      "step": 38770
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 4.641083333333334e-05,
+      "loss": 0.0059,
+      "step": 38780
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 4.640990740740741e-05,
+      "loss": 0.0045,
+      "step": 38790
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 4.640898148148148e-05,
+      "loss": 0.003,
+      "step": 38800
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 4.640805555555556e-05,
+      "loss": 0.0067,
+      "step": 38810
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 4.640712962962963e-05,
+      "loss": 0.0022,
+      "step": 38820
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 4.64062037037037e-05,
+      "loss": 0.0045,
+      "step": 38830
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 4.6405277777777777e-05,
+      "loss": 0.0037,
+      "step": 38840
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 4.640435185185185e-05,
+      "loss": 0.0018,
+      "step": 38850
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 4.6403425925925933e-05,
+      "loss": 0.0051,
+      "step": 38860
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 4.64025e-05,
+      "loss": 0.0027,
+      "step": 38870
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 4.640157407407408e-05,
+      "loss": 0.0058,
+      "step": 38880
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 4.640064814814815e-05,
+      "loss": 0.004,
+      "step": 38890
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 4.639972222222223e-05,
+      "loss": 0.0026,
+      "step": 38900
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 4.63987962962963e-05,
+      "loss": 0.0037,
+      "step": 38910
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 4.639787037037037e-05,
+      "loss": 0.0069,
+      "step": 38920
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 4.6396944444444445e-05,
+      "loss": 0.0062,
+      "step": 38930
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 4.639601851851852e-05,
+      "loss": 0.0043,
+      "step": 38940
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 4.6395092592592595e-05,
+      "loss": 0.0053,
+      "step": 38950
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 4.6394166666666664e-05,
+      "loss": 0.0042,
+      "step": 38960
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 4.6393240740740745e-05,
+      "loss": 0.0052,
+      "step": 38970
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 4.639231481481482e-05,
+      "loss": 0.0051,
+      "step": 38980
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 4.6391388888888896e-05,
+      "loss": 0.0037,
+      "step": 38990
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 4.6390462962962964e-05,
+      "loss": 0.0026,
+      "step": 39000
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 4.638953703703704e-05,
+      "loss": 0.0059,
+      "step": 39010
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 4.6388611111111114e-05,
+      "loss": 0.0068,
+      "step": 39020
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 4.638768518518519e-05,
+      "loss": 0.0028,
+      "step": 39030
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 4.638675925925926e-05,
+      "loss": 0.0019,
+      "step": 39040
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 4.638583333333333e-05,
+      "loss": 0.0032,
+      "step": 39050
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 4.638490740740741e-05,
+      "loss": 0.006,
+      "step": 39060
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 4.638398148148148e-05,
+      "loss": 0.004,
+      "step": 39070
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 4.638305555555556e-05,
+      "loss": 0.0038,
+      "step": 39080
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 4.638212962962963e-05,
+      "loss": 0.0046,
+      "step": 39090
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 4.638120370370371e-05,
+      "loss": 0.0038,
+      "step": 39100
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 4.638027777777778e-05,
+      "loss": 0.0028,
+      "step": 39110
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 4.637935185185186e-05,
+      "loss": 0.0071,
+      "step": 39120
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 4.6378425925925926e-05,
+      "loss": 0.0056,
+      "step": 39130
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 4.63775e-05,
+      "loss": 0.0033,
+      "step": 39140
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 4.6376574074074076e-05,
+      "loss": 0.0042,
+      "step": 39150
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 4.637564814814815e-05,
+      "loss": 0.0067,
+      "step": 39160
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 4.637472222222222e-05,
+      "loss": 0.009,
+      "step": 39170
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 4.63737962962963e-05,
+      "loss": 0.0031,
+      "step": 39180
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 4.6372870370370376e-05,
+      "loss": 0.0057,
+      "step": 39190
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 4.637194444444445e-05,
+      "loss": 0.003,
+      "step": 39200
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 4.637101851851852e-05,
+      "loss": 0.004,
+      "step": 39210
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 4.6370092592592595e-05,
+      "loss": 0.0036,
+      "step": 39220
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 4.636916666666667e-05,
+      "loss": 0.0056,
+      "step": 39230
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 4.6368240740740745e-05,
+      "loss": 0.0054,
+      "step": 39240
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 4.636731481481481e-05,
+      "loss": 0.0043,
+      "step": 39250
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 4.636638888888889e-05,
+      "loss": 0.0053,
+      "step": 39260
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 4.636546296296296e-05,
+      "loss": 0.0052,
+      "step": 39270
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 4.636453703703704e-05,
+      "loss": 0.0049,
+      "step": 39280
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 4.636361111111111e-05,
+      "loss": 0.004,
+      "step": 39290
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 4.636268518518519e-05,
+      "loss": 0.0067,
+      "step": 39300
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 4.6361759259259263e-05,
+      "loss": 0.0052,
+      "step": 39310
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 4.636083333333334e-05,
+      "loss": 0.0049,
+      "step": 39320
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 4.6359907407407414e-05,
+      "loss": 0.0067,
+      "step": 39330
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 4.635898148148148e-05,
+      "loss": 0.0027,
+      "step": 39340
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 4.635805555555556e-05,
+      "loss": 0.0062,
+      "step": 39350
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 4.635712962962963e-05,
+      "loss": 0.0039,
+      "step": 39360
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 4.635620370370371e-05,
+      "loss": 0.0047,
+      "step": 39370
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 4.6355277777777775e-05,
+      "loss": 0.0044,
+      "step": 39380
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 4.635435185185185e-05,
+      "loss": 0.0046,
+      "step": 39390
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 4.635342592592593e-05,
+      "loss": 0.0023,
+      "step": 39400
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 4.635250000000001e-05,
+      "loss": 0.0043,
+      "step": 39410
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 4.6351574074074076e-05,
+      "loss": 0.0036,
+      "step": 39420
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 4.635064814814815e-05,
+      "loss": 0.0033,
+      "step": 39430
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 4.6349722222222226e-05,
+      "loss": 0.005,
+      "step": 39440
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 4.63487962962963e-05,
+      "loss": 0.0041,
+      "step": 39450
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 4.634787037037037e-05,
+      "loss": 0.0023,
+      "step": 39460
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 4.6346944444444444e-05,
+      "loss": 0.0053,
+      "step": 39470
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 4.634601851851852e-05,
+      "loss": 0.0035,
+      "step": 39480
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 4.6345092592592594e-05,
+      "loss": 0.0047,
+      "step": 39490
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 4.634416666666667e-05,
+      "loss": 0.0038,
+      "step": 39500
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 4.6343240740740744e-05,
+      "loss": 0.0039,
+      "step": 39510
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 4.634231481481482e-05,
+      "loss": 0.0042,
+      "step": 39520
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 4.6341388888888894e-05,
+      "loss": 0.0048,
+      "step": 39530
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 4.634046296296297e-05,
+      "loss": 0.0065,
+      "step": 39540
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 4.633953703703704e-05,
+      "loss": 0.0071,
+      "step": 39550
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 4.633861111111111e-05,
+      "loss": 0.0076,
+      "step": 39560
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 4.633768518518519e-05,
+      "loss": 0.0031,
+      "step": 39570
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 4.633675925925926e-05,
+      "loss": 0.0045,
+      "step": 39580
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 4.633583333333333e-05,
+      "loss": 0.0049,
+      "step": 39590
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 4.6334907407407406e-05,
+      "loss": 0.0029,
+      "step": 39600
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 4.633398148148148e-05,
+      "loss": 0.009,
+      "step": 39610
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 4.633305555555556e-05,
+      "loss": 0.0064,
+      "step": 39620
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 4.633212962962963e-05,
+      "loss": 0.0036,
+      "step": 39630
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 4.6331203703703706e-05,
+      "loss": 0.0065,
+      "step": 39640
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 4.633027777777778e-05,
+      "loss": 0.0031,
+      "step": 39650
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 4.6329351851851857e-05,
+      "loss": 0.0046,
+      "step": 39660
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 4.6328425925925925e-05,
+      "loss": 0.0038,
+      "step": 39670
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 4.63275e-05,
+      "loss": 0.005,
+      "step": 39680
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 4.6326574074074075e-05,
+      "loss": 0.0055,
+      "step": 39690
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 4.632564814814815e-05,
+      "loss": 0.0051,
+      "step": 39700
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 4.6324722222222225e-05,
+      "loss": 0.0054,
+      "step": 39710
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 4.63237962962963e-05,
+      "loss": 0.0054,
+      "step": 39720
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 4.6322870370370375e-05,
+      "loss": 0.0059,
+      "step": 39730
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 4.632194444444445e-05,
+      "loss": 0.0081,
+      "step": 39740
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 4.6321018518518525e-05,
+      "loss": 0.005,
+      "step": 39750
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 4.6320092592592594e-05,
+      "loss": 0.0054,
+      "step": 39760
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 4.631916666666667e-05,
+      "loss": 0.0038,
+      "step": 39770
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 4.6318240740740744e-05,
+      "loss": 0.0035,
+      "step": 39780
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 4.631731481481482e-05,
+      "loss": 0.005,
+      "step": 39790
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 4.631638888888889e-05,
+      "loss": 0.0035,
+      "step": 39800
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 4.631546296296296e-05,
+      "loss": 0.0033,
+      "step": 39810
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 4.631453703703704e-05,
+      "loss": 0.0041,
+      "step": 39820
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 4.631361111111112e-05,
+      "loss": 0.0076,
+      "step": 39830
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 4.631268518518519e-05,
+      "loss": 0.0036,
+      "step": 39840
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 4.631175925925926e-05,
+      "loss": 0.0039,
+      "step": 39850
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 4.631083333333334e-05,
+      "loss": 0.0042,
+      "step": 39860
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 4.630990740740741e-05,
+      "loss": 0.003,
+      "step": 39870
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 4.630898148148148e-05,
+      "loss": 0.0041,
+      "step": 39880
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 4.6308055555555556e-05,
+      "loss": 0.0042,
+      "step": 39890
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 4.630712962962963e-05,
+      "loss": 0.0043,
+      "step": 39900
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 4.6306203703703706e-05,
+      "loss": 0.0042,
+      "step": 39910
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 4.630527777777778e-05,
+      "loss": 0.0057,
+      "step": 39920
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 4.630435185185185e-05,
+      "loss": 0.0036,
+      "step": 39930
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 4.630342592592593e-05,
+      "loss": 0.0052,
+      "step": 39940
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 4.6302500000000006e-05,
+      "loss": 0.0045,
+      "step": 39950
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 4.630157407407408e-05,
+      "loss": 0.0038,
+      "step": 39960
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 4.630064814814815e-05,
+      "loss": 0.0077,
+      "step": 39970
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 4.6299722222222224e-05,
+      "loss": 0.0035,
+      "step": 39980
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 4.62987962962963e-05,
+      "loss": 0.0066,
+      "step": 39990
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 4.6297870370370375e-05,
+      "loss": 0.0033,
+      "step": 40000
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 4.629694444444444e-05,
+      "loss": 0.0051,
+      "step": 40010
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 4.629601851851852e-05,
+      "loss": 0.0024,
+      "step": 40020
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 4.629509259259259e-05,
+      "loss": 0.0078,
+      "step": 40030
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 4.629416666666667e-05,
+      "loss": 0.0041,
+      "step": 40040
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 4.629324074074074e-05,
+      "loss": 0.0062,
+      "step": 40050
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 4.629231481481482e-05,
+      "loss": 0.0043,
+      "step": 40060
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 4.629138888888889e-05,
+      "loss": 0.0057,
+      "step": 40070
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 4.629046296296297e-05,
+      "loss": 0.0037,
+      "step": 40080
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 4.6289537037037036e-05,
+      "loss": 0.0029,
+      "step": 40090
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 4.628861111111111e-05,
+      "loss": 0.0038,
+      "step": 40100
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 4.6287685185185187e-05,
+      "loss": 0.0049,
+      "step": 40110
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 4.628675925925926e-05,
+      "loss": 0.0047,
+      "step": 40120
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 4.628583333333334e-05,
+      "loss": 0.0046,
+      "step": 40130
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 4.6284907407407405e-05,
+      "loss": 0.0052,
+      "step": 40140
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 4.628398148148148e-05,
+      "loss": 0.0034,
+      "step": 40150
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 4.628305555555556e-05,
+      "loss": 0.0043,
+      "step": 40160
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 4.628212962962964e-05,
+      "loss": 0.0056,
+      "step": 40170
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 4.6281203703703705e-05,
+      "loss": 0.0039,
+      "step": 40180
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 4.628027777777778e-05,
+      "loss": 0.0054,
+      "step": 40190
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 4.6279351851851855e-05,
+      "loss": 0.0044,
+      "step": 40200
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 4.627842592592593e-05,
+      "loss": 0.0049,
+      "step": 40210
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 4.62775e-05,
+      "loss": 0.0062,
+      "step": 40220
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 4.6276574074074074e-05,
+      "loss": 0.0032,
+      "step": 40230
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 4.627564814814815e-05,
+      "loss": 0.003,
+      "step": 40240
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 4.6274722222222224e-05,
+      "loss": 0.0039,
+      "step": 40250
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 4.62737962962963e-05,
+      "loss": 0.0034,
+      "step": 40260
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 4.6272870370370374e-05,
+      "loss": 0.0048,
+      "step": 40270
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 4.627194444444445e-05,
+      "loss": 0.0057,
+      "step": 40280
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 4.6271018518518524e-05,
+      "loss": 0.004,
+      "step": 40290
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 4.627009259259259e-05,
+      "loss": 0.0035,
+      "step": 40300
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 4.626916666666667e-05,
+      "loss": 0.0044,
+      "step": 40310
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 4.626824074074074e-05,
+      "loss": 0.0067,
+      "step": 40320
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 4.626731481481482e-05,
+      "loss": 0.0048,
+      "step": 40330
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 4.626638888888889e-05,
+      "loss": 0.0061,
+      "step": 40340
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 4.626546296296296e-05,
+      "loss": 0.0051,
+      "step": 40350
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 4.6264537037037036e-05,
+      "loss": 0.0039,
+      "step": 40360
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 4.626361111111112e-05,
+      "loss": 0.0056,
+      "step": 40370
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 4.626268518518519e-05,
+      "loss": 0.0041,
+      "step": 40380
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 4.626175925925926e-05,
+      "loss": 0.0057,
+      "step": 40390
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 4.6260833333333336e-05,
+      "loss": 0.0055,
+      "step": 40400
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 4.625990740740741e-05,
+      "loss": 0.0035,
+      "step": 40410
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 4.6258981481481486e-05,
+      "loss": 0.0032,
+      "step": 40420
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 4.6258055555555554e-05,
+      "loss": 0.0057,
+      "step": 40430
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 4.625712962962963e-05,
+      "loss": 0.0065,
+      "step": 40440
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 4.6256203703703705e-05,
+      "loss": 0.0038,
+      "step": 40450
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 4.625527777777778e-05,
+      "loss": 0.0042,
+      "step": 40460
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 4.6254351851851855e-05,
+      "loss": 0.0057,
+      "step": 40470
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 4.625342592592593e-05,
+      "loss": 0.0035,
+      "step": 40480
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 4.6252500000000005e-05,
+      "loss": 0.0052,
+      "step": 40490
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 4.625157407407408e-05,
+      "loss": 0.0032,
+      "step": 40500
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 4.625064814814815e-05,
+      "loss": 0.0048,
+      "step": 40510
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 4.624972222222222e-05,
+      "loss": 0.0057,
+      "step": 40520
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.62487962962963e-05,
+      "loss": 0.0083,
+      "step": 40530
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.624787037037037e-05,
+      "loss": 0.0039,
+      "step": 40540
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.624694444444445e-05,
+      "loss": 0.0053,
+      "step": 40550
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.624601851851852e-05,
+      "loss": 0.0033,
+      "step": 40560
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.624509259259259e-05,
+      "loss": 0.0061,
+      "step": 40570
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 4.624416666666667e-05,
+      "loss": 0.0058,
+      "step": 40580
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.624324074074075e-05,
+      "loss": 0.0057,
+      "step": 40590
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.624231481481482e-05,
+      "loss": 0.0057,
+      "step": 40600
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.624138888888889e-05,
+      "loss": 0.0025,
+      "step": 40610
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.624046296296297e-05,
+      "loss": 0.0039,
+      "step": 40620
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 4.623953703703704e-05,
+      "loss": 0.0032,
+      "step": 40630
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.623861111111111e-05,
+      "loss": 0.0027,
+      "step": 40640
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.6237685185185185e-05,
+      "loss": 0.0029,
+      "step": 40650
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.623675925925926e-05,
+      "loss": 0.0051,
+      "step": 40660
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.6235833333333335e-05,
+      "loss": 0.0039,
+      "step": 40670
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 4.623490740740741e-05,
+      "loss": 0.0045,
+      "step": 40680
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.6233981481481486e-05,
+      "loss": 0.0072,
+      "step": 40690
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.623305555555556e-05,
+      "loss": 0.0068,
+      "step": 40700
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.6232129629629636e-05,
+      "loss": 0.006,
+      "step": 40710
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.6231203703703704e-05,
+      "loss": 0.0037,
+      "step": 40720
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.623027777777778e-05,
+      "loss": 0.0024,
+      "step": 40730
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 4.6229351851851854e-05,
+      "loss": 0.004,
+      "step": 40740
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.622842592592593e-05,
+      "loss": 0.0036,
+      "step": 40750
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.6227500000000004e-05,
+      "loss": 0.0063,
+      "step": 40760
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.622666666666667e-05,
+      "loss": 0.0066,
+      "step": 40770
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.6225740740740744e-05,
+      "loss": 0.0063,
+      "step": 40780
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 4.622481481481481e-05,
+      "loss": 0.0083,
+      "step": 40790
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.622388888888889e-05,
+      "loss": 0.008,
+      "step": 40800
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.622296296296297e-05,
+      "loss": 0.005,
+      "step": 40810
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.6222037037037044e-05,
+      "loss": 0.0045,
+      "step": 40820
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.622111111111111e-05,
+      "loss": 0.0053,
+      "step": 40830
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.622018518518519e-05,
+      "loss": 0.0044,
+      "step": 40840
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 4.621925925925926e-05,
+      "loss": 0.0062,
+      "step": 40850
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.621833333333334e-05,
+      "loss": 0.0044,
+      "step": 40860
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.6217407407407406e-05,
+      "loss": 0.0057,
+      "step": 40870
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.621648148148148e-05,
+      "loss": 0.0038,
+      "step": 40880
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.6215555555555556e-05,
+      "loss": 0.0058,
+      "step": 40890
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 4.621462962962963e-05,
+      "loss": 0.0038,
+      "step": 40900
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.6213703703703706e-05,
+      "loss": 0.0061,
+      "step": 40910
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.621277777777778e-05,
+      "loss": 0.0056,
+      "step": 40920
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.6211851851851856e-05,
+      "loss": 0.0045,
+      "step": 40930
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.621092592592593e-05,
+      "loss": 0.0088,
+      "step": 40940
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 4.6210000000000006e-05,
+      "loss": 0.0045,
+      "step": 40950
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.6209074074074075e-05,
+      "loss": 0.0053,
+      "step": 40960
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.620814814814815e-05,
+      "loss": 0.0065,
+      "step": 40970
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.6207222222222225e-05,
+      "loss": 0.0082,
+      "step": 40980
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.62062962962963e-05,
+      "loss": 0.0066,
+      "step": 40990
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.620537037037037e-05,
+      "loss": 0.0035,
+      "step": 41000
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 4.620444444444444e-05,
+      "loss": 0.0048,
+      "step": 41010
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.620351851851852e-05,
+      "loss": 0.0056,
+      "step": 41020
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.62025925925926e-05,
+      "loss": 0.0034,
+      "step": 41030
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.620166666666667e-05,
+      "loss": 0.0044,
+      "step": 41040
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.6200740740740743e-05,
+      "loss": 0.0066,
+      "step": 41050
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 4.619981481481482e-05,
+      "loss": 0.0047,
+      "step": 41060
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 4.6198888888888894e-05,
+      "loss": 0.0057,
+      "step": 41070
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 4.619796296296296e-05,
+      "loss": 0.0029,
+      "step": 41080
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 4.619703703703704e-05,
+      "loss": 0.0047,
+      "step": 41090
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 4.619611111111111e-05,
+      "loss": 0.0044,
+      "step": 41100
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 4.619518518518519e-05,
+      "loss": 0.004,
+      "step": 41110
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 4.619425925925926e-05,
+      "loss": 0.0026,
+      "step": 41120
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 4.619333333333333e-05,
+      "loss": 0.0029,
+      "step": 41130
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 4.619240740740741e-05,
+      "loss": 0.0068,
+      "step": 41140
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 4.619148148148149e-05,
+      "loss": 0.0059,
+      "step": 41150
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 4.619055555555556e-05,
+      "loss": 0.0061,
+      "step": 41160
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 4.618962962962963e-05,
+      "loss": 0.0062,
+      "step": 41170
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 4.6188703703703706e-05,
+      "loss": 0.0044,
+      "step": 41180
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 4.618777777777778e-05,
+      "loss": 0.0036,
+      "step": 41190
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 4.6186851851851856e-05,
+      "loss": 0.0039,
+      "step": 41200
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 4.6185925925925924e-05,
+      "loss": 0.0051,
+      "step": 41210
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 4.6185e-05,
+      "loss": 0.004,
+      "step": 41220
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 4.6184074074074074e-05,
+      "loss": 0.0054,
+      "step": 41230
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 4.618314814814815e-05,
+      "loss": 0.0049,
+      "step": 41240
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 4.6182222222222224e-05,
+      "loss": 0.0022,
+      "step": 41250
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 4.61812962962963e-05,
+      "loss": 0.0054,
+      "step": 41260
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 4.6180370370370374e-05,
+      "loss": 0.006,
+      "step": 41270
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 4.617944444444445e-05,
+      "loss": 0.0042,
+      "step": 41280
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 4.617851851851852e-05,
+      "loss": 0.0045,
+      "step": 41290
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 4.617759259259259e-05,
+      "loss": 0.0052,
+      "step": 41300
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 4.617666666666667e-05,
+      "loss": 0.0083,
+      "step": 41310
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 4.617574074074074e-05,
+      "loss": 0.0036,
+      "step": 41320
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 4.617481481481482e-05,
+      "loss": 0.0068,
+      "step": 41330
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 4.6173888888888886e-05,
+      "loss": 0.0048,
+      "step": 41340
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 4.617296296296297e-05,
+      "loss": 0.0023,
+      "step": 41350
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 4.617203703703704e-05,
+      "loss": 0.0052,
+      "step": 41360
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 4.617111111111112e-05,
+      "loss": 0.0072,
+      "step": 41370
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 4.6170185185185186e-05,
+      "loss": 0.005,
+      "step": 41380
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 4.616925925925926e-05,
+      "loss": 0.0057,
+      "step": 41390
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 4.6168333333333337e-05,
+      "loss": 0.0075,
+      "step": 41400
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 4.616740740740741e-05,
+      "loss": 0.0047,
+      "step": 41410
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 4.616648148148148e-05,
+      "loss": 0.0062,
+      "step": 41420
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 4.6165555555555555e-05,
+      "loss": 0.0055,
+      "step": 41430
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 4.616462962962963e-05,
+      "loss": 0.0053,
+      "step": 41440
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 4.6163703703703705e-05,
+      "loss": 0.0062,
+      "step": 41450
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 4.616277777777778e-05,
+      "loss": 0.0046,
+      "step": 41460
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 4.6161851851851855e-05,
+      "loss": 0.0025,
+      "step": 41470
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 4.616092592592593e-05,
+      "loss": 0.0059,
+      "step": 41480
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 4.6160000000000005e-05,
+      "loss": 0.0049,
+      "step": 41490
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 4.6159074074074074e-05,
+      "loss": 0.0056,
+      "step": 41500
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 4.615814814814815e-05,
+      "loss": 0.0052,
+      "step": 41510
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 4.6157222222222224e-05,
+      "loss": 0.004,
+      "step": 41520
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 4.61562962962963e-05,
+      "loss": 0.0048,
+      "step": 41530
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 4.6155370370370374e-05,
+      "loss": 0.0044,
+      "step": 41540
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 4.615444444444444e-05,
+      "loss": 0.0054,
+      "step": 41550
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 4.615351851851852e-05,
+      "loss": 0.006,
+      "step": 41560
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 4.61525925925926e-05,
+      "loss": 0.0039,
+      "step": 41570
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 4.6151666666666674e-05,
+      "loss": 0.0061,
+      "step": 41580
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 4.615074074074074e-05,
+      "loss": 0.0048,
+      "step": 41590
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 4.614981481481482e-05,
+      "loss": 0.0064,
+      "step": 41600
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 4.614888888888889e-05,
+      "loss": 0.0033,
+      "step": 41610
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 4.614796296296297e-05,
+      "loss": 0.0055,
+      "step": 41620
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 4.6147037037037036e-05,
+      "loss": 0.0113,
+      "step": 41630
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 4.614611111111111e-05,
+      "loss": 0.0038,
+      "step": 41640
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 4.6145185185185186e-05,
+      "loss": 0.0066,
+      "step": 41650
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 4.614425925925926e-05,
+      "loss": 0.0027,
+      "step": 41660
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 4.6143333333333336e-05,
+      "loss": 0.0028,
+      "step": 41670
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 4.614240740740741e-05,
+      "loss": 0.0046,
+      "step": 41680
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 4.6141481481481486e-05,
+      "loss": 0.0047,
+      "step": 41690
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 4.614055555555556e-05,
+      "loss": 0.0041,
+      "step": 41700
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 4.613962962962963e-05,
+      "loss": 0.0041,
+      "step": 41710
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 4.6138703703703704e-05,
+      "loss": 0.0039,
+      "step": 41720
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 4.613777777777778e-05,
+      "loss": 0.0049,
+      "step": 41730
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 4.6136851851851855e-05,
+      "loss": 0.0047,
+      "step": 41740
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 4.613592592592593e-05,
+      "loss": 0.0035,
+      "step": 41750
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 4.6135e-05,
+      "loss": 0.0035,
+      "step": 41760
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 4.613407407407407e-05,
+      "loss": 0.0056,
+      "step": 41770
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 4.6133148148148155e-05,
+      "loss": 0.0031,
+      "step": 41780
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 4.613222222222223e-05,
+      "loss": 0.002,
+      "step": 41790
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 4.61312962962963e-05,
+      "loss": 0.0045,
+      "step": 41800
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 4.613037037037037e-05,
+      "loss": 0.0054,
+      "step": 41810
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 4.612944444444445e-05,
+      "loss": 0.004,
+      "step": 41820
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 4.612851851851852e-05,
+      "loss": 0.0051,
+      "step": 41830
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 4.612759259259259e-05,
+      "loss": 0.0054,
+      "step": 41840
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 4.612666666666667e-05,
+      "loss": 0.0074,
+      "step": 41850
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 4.612574074074074e-05,
+      "loss": 0.0048,
+      "step": 41860
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 4.612481481481482e-05,
+      "loss": 0.0033,
+      "step": 41870
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 4.612388888888889e-05,
+      "loss": 0.0028,
+      "step": 41880
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 4.612296296296297e-05,
+      "loss": 0.0029,
+      "step": 41890
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 4.612203703703704e-05,
+      "loss": 0.0038,
+      "step": 41900
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 4.612111111111112e-05,
+      "loss": 0.0066,
+      "step": 41910
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 4.6120185185185185e-05,
+      "loss": 0.0072,
+      "step": 41920
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 4.611925925925926e-05,
+      "loss": 0.0049,
+      "step": 41930
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 4.6118333333333335e-05,
+      "loss": 0.0058,
+      "step": 41940
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 4.611740740740741e-05,
+      "loss": 0.0058,
+      "step": 41950
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 4.6116481481481485e-05,
+      "loss": 0.0031,
+      "step": 41960
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 4.6115555555555554e-05,
+      "loss": 0.0029,
+      "step": 41970
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 4.611462962962963e-05,
+      "loss": 0.0041,
+      "step": 41980
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 4.6113703703703704e-05,
+      "loss": 0.0053,
+      "step": 41990
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 4.6112777777777786e-05,
+      "loss": 0.0079,
+      "step": 42000
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 4.6111851851851854e-05,
+      "loss": 0.0046,
+      "step": 42010
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 4.611092592592593e-05,
+      "loss": 0.0063,
+      "step": 42020
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 4.6110000000000004e-05,
+      "loss": 0.0033,
+      "step": 42030
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 4.610907407407408e-05,
+      "loss": 0.0052,
+      "step": 42040
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 4.610814814814815e-05,
+      "loss": 0.0052,
+      "step": 42050
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 4.610722222222222e-05,
+      "loss": 0.0056,
+      "step": 42060
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 4.61062962962963e-05,
+      "loss": 0.0034,
+      "step": 42070
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 4.610537037037037e-05,
+      "loss": 0.0042,
+      "step": 42080
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 4.610444444444445e-05,
+      "loss": 0.0039,
+      "step": 42090
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 4.6103518518518516e-05,
+      "loss": 0.0044,
+      "step": 42100
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 4.61025925925926e-05,
+      "loss": 0.0054,
+      "step": 42110
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 4.610166666666667e-05,
+      "loss": 0.0028,
+      "step": 42120
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 4.610074074074074e-05,
+      "loss": 0.0082,
+      "step": 42130
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 4.6099814814814816e-05,
+      "loss": 0.0054,
+      "step": 42140
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 4.609888888888889e-05,
+      "loss": 0.0028,
+      "step": 42150
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 4.6097962962962966e-05,
+      "loss": 0.0029,
+      "step": 42160
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 4.609703703703704e-05,
+      "loss": 0.005,
+      "step": 42170
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 4.609611111111111e-05,
+      "loss": 0.0045,
+      "step": 42180
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 4.6095185185185185e-05,
+      "loss": 0.0037,
+      "step": 42190
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 4.609425925925926e-05,
+      "loss": 0.0062,
+      "step": 42200
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 4.6093333333333335e-05,
+      "loss": 0.0067,
+      "step": 42210
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 4.609240740740741e-05,
+      "loss": 0.0037,
+      "step": 42220
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 4.6091481481481485e-05,
+      "loss": 0.0042,
+      "step": 42230
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 4.609055555555556e-05,
+      "loss": 0.0081,
+      "step": 42240
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 4.6089629629629635e-05,
+      "loss": 0.004,
+      "step": 42250
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 4.60887037037037e-05,
+      "loss": 0.0024,
+      "step": 42260
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 4.608777777777778e-05,
+      "loss": 0.0046,
+      "step": 42270
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 4.608685185185185e-05,
+      "loss": 0.0052,
+      "step": 42280
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 4.608592592592593e-05,
+      "loss": 0.0053,
+      "step": 42290
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 4.6085000000000003e-05,
+      "loss": 0.0044,
+      "step": 42300
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 4.608407407407407e-05,
+      "loss": 0.0033,
+      "step": 42310
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 4.6083148148148154e-05,
+      "loss": 0.0064,
+      "step": 42320
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 4.608222222222223e-05,
+      "loss": 0.0034,
+      "step": 42330
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 4.60812962962963e-05,
+      "loss": 0.0053,
+      "step": 42340
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 4.608037037037037e-05,
+      "loss": 0.0083,
+      "step": 42350
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 4.607944444444445e-05,
+      "loss": 0.0054,
+      "step": 42360
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 4.607851851851852e-05,
+      "loss": 0.0054,
+      "step": 42370
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 4.60775925925926e-05,
+      "loss": 0.005,
+      "step": 42380
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 4.6076666666666665e-05,
+      "loss": 0.004,
+      "step": 42390
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 4.607574074074074e-05,
+      "loss": 0.0035,
+      "step": 42400
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 4.6074814814814816e-05,
+      "loss": 0.0038,
+      "step": 42410
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 4.607388888888889e-05,
+      "loss": 0.0027,
+      "step": 42420
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 4.6072962962962966e-05,
+      "loss": 0.0055,
+      "step": 42430
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 4.607203703703704e-05,
+      "loss": 0.0052,
+      "step": 42440
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 4.6071111111111116e-05,
+      "loss": 0.0027,
+      "step": 42450
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 4.607018518518519e-05,
+      "loss": 0.0044,
+      "step": 42460
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 4.606925925925926e-05,
+      "loss": 0.0042,
+      "step": 42470
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 4.6068333333333334e-05,
+      "loss": 0.0054,
+      "step": 42480
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 4.606740740740741e-05,
+      "loss": 0.0037,
+      "step": 42490
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 4.6066481481481484e-05,
+      "loss": 0.0038,
+      "step": 42500
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 4.606555555555556e-05,
+      "loss": 0.0044,
+      "step": 42510
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 4.606462962962963e-05,
+      "loss": 0.0055,
+      "step": 42520
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 4.60637037037037e-05,
+      "loss": 0.0059,
+      "step": 42530
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 4.6062777777777784e-05,
+      "loss": 0.0044,
+      "step": 42540
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 4.606185185185185e-05,
+      "loss": 0.005,
+      "step": 42550
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 4.606092592592593e-05,
+      "loss": 0.0045,
+      "step": 42560
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 4.606e-05,
+      "loss": 0.0061,
+      "step": 42570
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 4.605907407407408e-05,
+      "loss": 0.0048,
+      "step": 42580
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 4.605814814814815e-05,
+      "loss": 0.0062,
+      "step": 42590
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 4.605722222222222e-05,
+      "loss": 0.0042,
+      "step": 42600
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 4.6056296296296296e-05,
+      "loss": 0.0073,
+      "step": 42610
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 4.605537037037037e-05,
+      "loss": 0.0057,
+      "step": 42620
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 4.6054444444444446e-05,
+      "loss": 0.0029,
+      "step": 42630
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 4.6053518518518515e-05,
+      "loss": 0.0042,
+      "step": 42640
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 4.6052592592592597e-05,
+      "loss": 0.0069,
+      "step": 42650
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 4.605166666666667e-05,
+      "loss": 0.0064,
+      "step": 42660
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 4.605074074074075e-05,
+      "loss": 0.0033,
+      "step": 42670
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 4.6049814814814815e-05,
+      "loss": 0.005,
+      "step": 42680
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 4.604888888888889e-05,
+      "loss": 0.0054,
+      "step": 42690
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 4.6047962962962965e-05,
+      "loss": 0.0035,
+      "step": 42700
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 4.604703703703704e-05,
+      "loss": 0.0041,
+      "step": 42710
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 4.6046111111111115e-05,
+      "loss": 0.0052,
+      "step": 42720
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 4.6045185185185183e-05,
+      "loss": 0.0041,
+      "step": 42730
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 4.604425925925926e-05,
+      "loss": 0.0031,
+      "step": 42740
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 4.6043333333333334e-05,
+      "loss": 0.005,
+      "step": 42750
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 4.604240740740741e-05,
+      "loss": 0.0047,
+      "step": 42760
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 4.6041481481481484e-05,
+      "loss": 0.0054,
+      "step": 42770
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 4.604055555555556e-05,
+      "loss": 0.0037,
+      "step": 42780
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 4.6039629629629634e-05,
+      "loss": 0.0055,
+      "step": 42790
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 4.603870370370371e-05,
+      "loss": 0.0057,
+      "step": 42800
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 4.603777777777778e-05,
+      "loss": 0.0058,
+      "step": 42810
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 4.603685185185185e-05,
+      "loss": 0.0079,
+      "step": 42820
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 4.603601851851852e-05,
+      "loss": 0.0048,
+      "step": 42830
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 4.603509259259259e-05,
+      "loss": 0.0052,
+      "step": 42840
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 4.603416666666667e-05,
+      "loss": 0.0033,
+      "step": 42850
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 4.603324074074074e-05,
+      "loss": 0.0036,
+      "step": 42860
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 4.603231481481482e-05,
+      "loss": 0.0053,
+      "step": 42870
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 4.603138888888889e-05,
+      "loss": 0.0052,
+      "step": 42880
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 4.603046296296297e-05,
+      "loss": 0.0037,
+      "step": 42890
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 4.602953703703704e-05,
+      "loss": 0.0044,
+      "step": 42900
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 4.602861111111111e-05,
+      "loss": 0.0047,
+      "step": 42910
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 4.6027685185185186e-05,
+      "loss": 0.0059,
+      "step": 42920
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 4.602675925925926e-05,
+      "loss": 0.0039,
+      "step": 42930
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 4.6025833333333336e-05,
+      "loss": 0.0045,
+      "step": 42940
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 4.602490740740741e-05,
+      "loss": 0.0054,
+      "step": 42950
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.602398148148148e-05,
+      "loss": 0.0032,
+      "step": 42960
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.6023055555555554e-05,
+      "loss": 0.0064,
+      "step": 42970
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.6022129629629636e-05,
+      "loss": 0.0052,
+      "step": 42980
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.602120370370371e-05,
+      "loss": 0.0063,
+      "step": 42990
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.602027777777778e-05,
+      "loss": 0.0048,
+      "step": 43000
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 4.6019351851851854e-05,
+      "loss": 0.0029,
+      "step": 43010
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 4.601842592592593e-05,
+      "loss": 0.0035,
+      "step": 43020
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 4.6017500000000005e-05,
+      "loss": 0.0035,
+      "step": 43030
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 4.601657407407407e-05,
+      "loss": 0.0045,
+      "step": 43040
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 4.601564814814815e-05,
+      "loss": 0.0047,
+      "step": 43050
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 4.601472222222222e-05,
+      "loss": 0.0063,
+      "step": 43060
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 4.60137962962963e-05,
+      "loss": 0.0021,
+      "step": 43070
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 4.601287037037037e-05,
+      "loss": 0.0054,
+      "step": 43080
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 4.601194444444445e-05,
+      "loss": 0.0038,
+      "step": 43090
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 4.601101851851852e-05,
+      "loss": 0.0044,
+      "step": 43100
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 4.60100925925926e-05,
+      "loss": 0.0021,
+      "step": 43110
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 4.6009166666666666e-05,
+      "loss": 0.0049,
+      "step": 43120
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 4.600824074074074e-05,
+      "loss": 0.004,
+      "step": 43130
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 4.6007314814814817e-05,
+      "loss": 0.006,
+      "step": 43140
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 4.600638888888889e-05,
+      "loss": 0.0036,
+      "step": 43150
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 4.600546296296297e-05,
+      "loss": 0.0031,
+      "step": 43160
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 4.6004537037037035e-05,
+      "loss": 0.0052,
+      "step": 43170
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 4.600361111111111e-05,
+      "loss": 0.0049,
+      "step": 43180
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 4.6002685185185185e-05,
+      "loss": 0.0059,
+      "step": 43190
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 4.600175925925927e-05,
+      "loss": 0.006,
+      "step": 43200
+    },
+    {
+      "epoch": 8.0,
+      "eval_loss": 0.01357207540422678,
+      "eval_runtime": 153.5532,
+      "eval_samples_per_second": 25.984,
+      "eval_steps_per_second": 3.25,
+      "step": 43200
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 4.6000833333333335e-05,
+      "loss": 0.0061,
+      "step": 43210
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 4.599990740740741e-05,
+      "loss": 0.0033,
+      "step": 43220
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 4.5998981481481485e-05,
+      "loss": 0.0038,
+      "step": 43230
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 4.599805555555556e-05,
+      "loss": 0.0042,
+      "step": 43240
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 4.599712962962963e-05,
+      "loss": 0.0033,
+      "step": 43250
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 4.5996203703703704e-05,
+      "loss": 0.0032,
+      "step": 43260
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 4.599527777777778e-05,
+      "loss": 0.0042,
+      "step": 43270
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 4.5994351851851854e-05,
+      "loss": 0.0037,
+      "step": 43280
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 4.599342592592593e-05,
+      "loss": 0.0034,
+      "step": 43290
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 4.59925e-05,
+      "loss": 0.0054,
+      "step": 43300
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 4.599157407407408e-05,
+      "loss": 0.0032,
+      "step": 43310
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 4.5990648148148154e-05,
+      "loss": 0.0058,
+      "step": 43320
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 4.598972222222223e-05,
+      "loss": 0.0066,
+      "step": 43330
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 4.59887962962963e-05,
+      "loss": 0.0072,
+      "step": 43340
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 4.598787037037037e-05,
+      "loss": 0.005,
+      "step": 43350
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 4.598694444444445e-05,
+      "loss": 0.0035,
+      "step": 43360
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 4.598601851851852e-05,
+      "loss": 0.0033,
+      "step": 43370
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 4.598509259259259e-05,
+      "loss": 0.0035,
+      "step": 43380
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 4.5984166666666666e-05,
+      "loss": 0.0046,
+      "step": 43390
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 4.598324074074074e-05,
+      "loss": 0.0038,
+      "step": 43400
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 4.598231481481482e-05,
+      "loss": 0.0033,
+      "step": 43410
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 4.598138888888889e-05,
+      "loss": 0.0033,
+      "step": 43420
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 4.5980462962962966e-05,
+      "loss": 0.005,
+      "step": 43430
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 4.597953703703704e-05,
+      "loss": 0.0027,
+      "step": 43440
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 4.5978611111111116e-05,
+      "loss": 0.004,
+      "step": 43450
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 4.5977685185185184e-05,
+      "loss": 0.0016,
+      "step": 43460
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 4.597675925925926e-05,
+      "loss": 0.0049,
+      "step": 43470
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 4.5975833333333335e-05,
+      "loss": 0.0042,
+      "step": 43480
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 4.597490740740741e-05,
+      "loss": 0.0042,
+      "step": 43490
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 4.5973981481481485e-05,
+      "loss": 0.004,
+      "step": 43500
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 4.597305555555555e-05,
+      "loss": 0.0051,
+      "step": 43510
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 4.5972129629629635e-05,
+      "loss": 0.0032,
+      "step": 43520
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 4.597120370370371e-05,
+      "loss": 0.0034,
+      "step": 43530
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 4.5970277777777785e-05,
+      "loss": 0.0048,
+      "step": 43540
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 4.596935185185185e-05,
+      "loss": 0.0034,
+      "step": 43550
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 4.596842592592593e-05,
+      "loss": 0.0065,
+      "step": 43560
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 4.59675e-05,
+      "loss": 0.0071,
+      "step": 43570
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 4.596657407407408e-05,
+      "loss": 0.0035,
+      "step": 43580
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 4.596564814814815e-05,
+      "loss": 0.0068,
+      "step": 43590
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 4.596472222222222e-05,
+      "loss": 0.0036,
+      "step": 43600
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 4.59637962962963e-05,
+      "loss": 0.0038,
+      "step": 43610
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 4.596287037037037e-05,
+      "loss": 0.0054,
+      "step": 43620
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 4.596194444444445e-05,
+      "loss": 0.0043,
+      "step": 43630
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 4.596101851851852e-05,
+      "loss": 0.0033,
+      "step": 43640
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 4.596018518518519e-05,
+      "loss": 0.006,
+      "step": 43650
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 4.595925925925926e-05,
+      "loss": 0.0037,
+      "step": 43660
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 4.595833333333334e-05,
+      "loss": 0.0046,
+      "step": 43670
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 4.595740740740741e-05,
+      "loss": 0.0023,
+      "step": 43680
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 4.595648148148149e-05,
+      "loss": 0.0039,
+      "step": 43690
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 4.5955555555555555e-05,
+      "loss": 0.0042,
+      "step": 43700
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 4.595462962962963e-05,
+      "loss": 0.0035,
+      "step": 43710
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 4.5953703703703705e-05,
+      "loss": 0.0064,
+      "step": 43720
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 4.595277777777778e-05,
+      "loss": 0.0024,
+      "step": 43730
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 4.595185185185185e-05,
+      "loss": 0.0058,
+      "step": 43740
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 4.595092592592593e-05,
+      "loss": 0.0033,
+      "step": 43750
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 4.5950000000000006e-05,
+      "loss": 0.0039,
+      "step": 43760
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 4.594907407407408e-05,
+      "loss": 0.0039,
+      "step": 43770
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 4.594814814814815e-05,
+      "loss": 0.0037,
+      "step": 43780
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 4.5947222222222224e-05,
+      "loss": 0.0037,
+      "step": 43790
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 4.59462962962963e-05,
+      "loss": 0.0038,
+      "step": 43800
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 4.5945370370370374e-05,
+      "loss": 0.0036,
+      "step": 43810
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 4.594444444444444e-05,
+      "loss": 0.0028,
+      "step": 43820
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 4.594351851851852e-05,
+      "loss": 0.0058,
+      "step": 43830
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 4.594259259259259e-05,
+      "loss": 0.0033,
+      "step": 43840
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 4.594166666666667e-05,
+      "loss": 0.0051,
+      "step": 43850
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 4.594074074074074e-05,
+      "loss": 0.0052,
+      "step": 43860
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 4.593981481481482e-05,
+      "loss": 0.0032,
+      "step": 43870
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 4.593888888888889e-05,
+      "loss": 0.0035,
+      "step": 43880
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 4.593796296296297e-05,
+      "loss": 0.0043,
+      "step": 43890
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 4.593703703703704e-05,
+      "loss": 0.0037,
+      "step": 43900
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 4.593611111111111e-05,
+      "loss": 0.0039,
+      "step": 43910
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 4.5935185185185186e-05,
+      "loss": 0.003,
+      "step": 43920
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 4.593425925925926e-05,
+      "loss": 0.0081,
+      "step": 43930
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 4.5933333333333336e-05,
+      "loss": 0.0043,
+      "step": 43940
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 4.5932407407407405e-05,
+      "loss": 0.003,
+      "step": 43950
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 4.5931481481481486e-05,
+      "loss": 0.0045,
+      "step": 43960
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 4.593055555555556e-05,
+      "loss": 0.0052,
+      "step": 43970
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 4.5929629629629636e-05,
+      "loss": 0.0057,
+      "step": 43980
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 4.5928703703703705e-05,
+      "loss": 0.0045,
+      "step": 43990
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 4.592777777777778e-05,
+      "loss": 0.0033,
+      "step": 44000
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 4.5926851851851855e-05,
+      "loss": 0.004,
+      "step": 44010
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 4.592592592592593e-05,
+      "loss": 0.0045,
+      "step": 44020
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 4.5925e-05,
+      "loss": 0.0062,
+      "step": 44030
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 4.592407407407407e-05,
+      "loss": 0.0062,
+      "step": 44040
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 4.592314814814815e-05,
+      "loss": 0.0027,
+      "step": 44050
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 4.592222222222222e-05,
+      "loss": 0.0051,
+      "step": 44060
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 4.59212962962963e-05,
+      "loss": 0.0018,
+      "step": 44070
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 4.5920370370370373e-05,
+      "loss": 0.006,
+      "step": 44080
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 4.591944444444445e-05,
+      "loss": 0.005,
+      "step": 44090
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 4.5918518518518524e-05,
+      "loss": 0.005,
+      "step": 44100
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 4.59175925925926e-05,
+      "loss": 0.003,
+      "step": 44110
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 4.591666666666667e-05,
+      "loss": 0.0045,
+      "step": 44120
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 4.591574074074074e-05,
+      "loss": 0.0038,
+      "step": 44130
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 4.591481481481482e-05,
+      "loss": 0.0049,
+      "step": 44140
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 4.591388888888889e-05,
+      "loss": 0.0046,
+      "step": 44150
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 4.591296296296296e-05,
+      "loss": 0.0031,
+      "step": 44160
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 4.5912037037037035e-05,
+      "loss": 0.0049,
+      "step": 44170
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 4.591111111111112e-05,
+      "loss": 0.0044,
+      "step": 44180
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 4.591018518518519e-05,
+      "loss": 0.004,
+      "step": 44190
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 4.590925925925926e-05,
+      "loss": 0.0065,
+      "step": 44200
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 4.5908333333333336e-05,
+      "loss": 0.0058,
+      "step": 44210
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 4.590740740740741e-05,
+      "loss": 0.0037,
+      "step": 44220
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 4.5906481481481486e-05,
+      "loss": 0.0039,
+      "step": 44230
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 4.5905555555555554e-05,
+      "loss": 0.0023,
+      "step": 44240
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 4.590462962962963e-05,
+      "loss": 0.0018,
+      "step": 44250
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 4.5903703703703704e-05,
+      "loss": 0.0047,
+      "step": 44260
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 4.590277777777778e-05,
+      "loss": 0.0065,
+      "step": 44270
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 4.5901851851851854e-05,
+      "loss": 0.0035,
+      "step": 44280
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 4.590092592592593e-05,
+      "loss": 0.0054,
+      "step": 44290
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 4.5900000000000004e-05,
+      "loss": 0.0051,
+      "step": 44300
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 4.589907407407408e-05,
+      "loss": 0.0027,
+      "step": 44310
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 4.5898148148148154e-05,
+      "loss": 0.0069,
+      "step": 44320
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 4.589722222222222e-05,
+      "loss": 0.0035,
+      "step": 44330
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 4.58962962962963e-05,
+      "loss": 0.0038,
+      "step": 44340
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 4.589537037037037e-05,
+      "loss": 0.0036,
+      "step": 44350
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 4.589444444444445e-05,
+      "loss": 0.0035,
+      "step": 44360
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 4.5893518518518516e-05,
+      "loss": 0.0046,
+      "step": 44370
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 4.589259259259259e-05,
+      "loss": 0.0056,
+      "step": 44380
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 4.5891666666666666e-05,
+      "loss": 0.0042,
+      "step": 44390
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 4.589074074074075e-05,
+      "loss": 0.0045,
+      "step": 44400
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 4.5889814814814816e-05,
+      "loss": 0.0042,
+      "step": 44410
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 4.588888888888889e-05,
+      "loss": 0.0038,
+      "step": 44420
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 4.5887962962962967e-05,
+      "loss": 0.0029,
+      "step": 44430
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 4.588703703703704e-05,
+      "loss": 0.0047,
+      "step": 44440
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 4.588611111111111e-05,
+      "loss": 0.0024,
+      "step": 44450
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 4.5885185185185185e-05,
+      "loss": 0.0047,
+      "step": 44460
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 4.588425925925926e-05,
+      "loss": 0.0105,
+      "step": 44470
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 4.5883333333333335e-05,
+      "loss": 0.0027,
+      "step": 44480
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 4.588240740740741e-05,
+      "loss": 0.0028,
+      "step": 44490
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 4.5881481481481485e-05,
+      "loss": 0.0046,
+      "step": 44500
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 4.588055555555556e-05,
+      "loss": 0.0059,
+      "step": 44510
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 4.5879629629629635e-05,
+      "loss": 0.0029,
+      "step": 44520
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 4.587870370370371e-05,
+      "loss": 0.0034,
+      "step": 44530
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 4.587777777777778e-05,
+      "loss": 0.0048,
+      "step": 44540
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 4.5876851851851854e-05,
+      "loss": 0.0055,
+      "step": 44550
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 4.587592592592593e-05,
+      "loss": 0.0068,
+      "step": 44560
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 4.5875000000000004e-05,
+      "loss": 0.0047,
+      "step": 44570
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 4.587407407407407e-05,
+      "loss": 0.0043,
+      "step": 44580
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 4.587314814814815e-05,
+      "loss": 0.0041,
+      "step": 44590
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 4.587222222222222e-05,
+      "loss": 0.005,
+      "step": 44600
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 4.5871296296296304e-05,
+      "loss": 0.0063,
+      "step": 44610
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 4.587037037037037e-05,
+      "loss": 0.0055,
+      "step": 44620
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 4.586944444444445e-05,
+      "loss": 0.0039,
+      "step": 44630
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 4.586851851851852e-05,
+      "loss": 0.005,
+      "step": 44640
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 4.58675925925926e-05,
+      "loss": 0.0042,
+      "step": 44650
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 4.5866666666666666e-05,
+      "loss": 0.0036,
+      "step": 44660
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 4.586574074074074e-05,
+      "loss": 0.0036,
+      "step": 44670
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 4.5864814814814816e-05,
+      "loss": 0.0046,
+      "step": 44680
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 4.586388888888889e-05,
+      "loss": 0.005,
+      "step": 44690
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 4.5862962962962966e-05,
+      "loss": 0.0047,
+      "step": 44700
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 4.5862037037037034e-05,
+      "loss": 0.0026,
+      "step": 44710
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 4.5861111111111116e-05,
+      "loss": 0.0054,
+      "step": 44720
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 4.586018518518519e-05,
+      "loss": 0.0055,
+      "step": 44730
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 4.5859259259259266e-05,
+      "loss": 0.0039,
+      "step": 44740
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 4.5858333333333334e-05,
+      "loss": 0.0046,
+      "step": 44750
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 4.585740740740741e-05,
+      "loss": 0.0027,
+      "step": 44760
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 4.5856481481481485e-05,
+      "loss": 0.0059,
+      "step": 44770
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 4.585555555555556e-05,
+      "loss": 0.0037,
+      "step": 44780
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 4.585462962962963e-05,
+      "loss": 0.0041,
+      "step": 44790
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 4.58537037037037e-05,
+      "loss": 0.0034,
+      "step": 44800
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 4.585277777777778e-05,
+      "loss": 0.0043,
+      "step": 44810
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 4.585185185185185e-05,
+      "loss": 0.0043,
+      "step": 44820
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 4.585092592592593e-05,
+      "loss": 0.0058,
+      "step": 44830
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 4.585e-05,
+      "loss": 0.0015,
+      "step": 44840
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 4.584907407407408e-05,
+      "loss": 0.0056,
+      "step": 44850
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 4.584814814814815e-05,
+      "loss": 0.0053,
+      "step": 44860
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 4.584722222222222e-05,
+      "loss": 0.0029,
+      "step": 44870
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 4.5846296296296297e-05,
+      "loss": 0.0063,
+      "step": 44880
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 4.584537037037037e-05,
+      "loss": 0.0035,
+      "step": 44890
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 4.584444444444445e-05,
+      "loss": 0.0044,
+      "step": 44900
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 4.584351851851852e-05,
+      "loss": 0.0034,
+      "step": 44910
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 4.584259259259259e-05,
+      "loss": 0.0041,
+      "step": 44920
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 4.5841666666666665e-05,
+      "loss": 0.0041,
+      "step": 44930
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 4.584074074074075e-05,
+      "loss": 0.0032,
+      "step": 44940
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 4.583981481481482e-05,
+      "loss": 0.005,
+      "step": 44950
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 4.583888888888889e-05,
+      "loss": 0.0035,
+      "step": 44960
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 4.5837962962962965e-05,
+      "loss": 0.0045,
+      "step": 44970
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 4.583703703703704e-05,
+      "loss": 0.003,
+      "step": 44980
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 4.5836111111111115e-05,
+      "loss": 0.004,
+      "step": 44990
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 4.5835185185185184e-05,
+      "loss": 0.0055,
+      "step": 45000
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 4.583425925925926e-05,
+      "loss": 0.0051,
+      "step": 45010
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 4.5833333333333334e-05,
+      "loss": 0.0023,
+      "step": 45020
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 4.583240740740741e-05,
+      "loss": 0.0057,
+      "step": 45030
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 4.5831481481481484e-05,
+      "loss": 0.0047,
+      "step": 45040
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 4.583055555555556e-05,
+      "loss": 0.0043,
+      "step": 45050
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 4.5829629629629634e-05,
+      "loss": 0.0025,
+      "step": 45060
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 4.582870370370371e-05,
+      "loss": 0.0031,
+      "step": 45070
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 4.582777777777778e-05,
+      "loss": 0.0036,
+      "step": 45080
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 4.582685185185185e-05,
+      "loss": 0.0038,
+      "step": 45090
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 4.582592592592593e-05,
+      "loss": 0.0039,
+      "step": 45100
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 4.5825e-05,
+      "loss": 0.0027,
+      "step": 45110
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.582407407407408e-05,
+      "loss": 0.0028,
+      "step": 45120
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.5823148148148146e-05,
+      "loss": 0.0062,
+      "step": 45130
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.582222222222222e-05,
+      "loss": 0.0042,
+      "step": 45140
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.58212962962963e-05,
+      "loss": 0.0032,
+      "step": 45150
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.582037037037038e-05,
+      "loss": 0.0018,
+      "step": 45160
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 4.5819444444444446e-05,
+      "loss": 0.0036,
+      "step": 45170
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 4.581851851851852e-05,
+      "loss": 0.0052,
+      "step": 45180
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 4.5817592592592596e-05,
+      "loss": 0.004,
+      "step": 45190
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 4.581666666666667e-05,
+      "loss": 0.003,
+      "step": 45200
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 4.581574074074074e-05,
+      "loss": 0.0031,
+      "step": 45210
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 4.5814814814814815e-05,
+      "loss": 0.0078,
+      "step": 45220
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.581388888888889e-05,
+      "loss": 0.0044,
+      "step": 45230
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.5812962962962965e-05,
+      "loss": 0.0039,
+      "step": 45240
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.581203703703704e-05,
+      "loss": 0.0042,
+      "step": 45250
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.5811111111111115e-05,
+      "loss": 0.0068,
+      "step": 45260
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 4.581018518518519e-05,
+      "loss": 0.0029,
+      "step": 45270
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.5809259259259265e-05,
+      "loss": 0.0043,
+      "step": 45280
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.580833333333333e-05,
+      "loss": 0.0028,
+      "step": 45290
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.580740740740741e-05,
+      "loss": 0.0038,
+      "step": 45300
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.580648148148148e-05,
+      "loss": 0.0072,
+      "step": 45310
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.580555555555556e-05,
+      "loss": 0.0046,
+      "step": 45320
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 4.5804629629629633e-05,
+      "loss": 0.003,
+      "step": 45330
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.58037037037037e-05,
+      "loss": 0.006,
+      "step": 45340
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.580277777777778e-05,
+      "loss": 0.006,
+      "step": 45350
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.580185185185185e-05,
+      "loss": 0.0035,
+      "step": 45360
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.5800925925925934e-05,
+      "loss": 0.004,
+      "step": 45370
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 4.58e-05,
+      "loss": 0.0048,
+      "step": 45380
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.579907407407408e-05,
+      "loss": 0.0036,
+      "step": 45390
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.579814814814815e-05,
+      "loss": 0.0039,
+      "step": 45400
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.579722222222223e-05,
+      "loss": 0.0051,
+      "step": 45410
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.5796296296296295e-05,
+      "loss": 0.0064,
+      "step": 45420
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.579537037037037e-05,
+      "loss": 0.0043,
+      "step": 45430
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 4.5794444444444446e-05,
+      "loss": 0.0038,
+      "step": 45440
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 4.579351851851852e-05,
+      "loss": 0.0049,
+      "step": 45450
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 4.5792592592592596e-05,
+      "loss": 0.005,
+      "step": 45460
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 4.579166666666667e-05,
+      "loss": 0.0033,
+      "step": 45470
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 4.5790740740740746e-05,
+      "loss": 0.0025,
+      "step": 45480
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 4.578981481481482e-05,
+      "loss": 0.0052,
+      "step": 45490
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.578888888888889e-05,
+      "loss": 0.0046,
+      "step": 45500
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.5787962962962964e-05,
+      "loss": 0.0041,
+      "step": 45510
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.578703703703704e-05,
+      "loss": 0.0068,
+      "step": 45520
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.5786111111111114e-05,
+      "loss": 0.0066,
+      "step": 45530
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 4.578518518518519e-05,
+      "loss": 0.0028,
+      "step": 45540
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.578425925925926e-05,
+      "loss": 0.0061,
+      "step": 45550
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.578333333333333e-05,
+      "loss": 0.0041,
+      "step": 45560
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.578240740740741e-05,
+      "loss": 0.0054,
+      "step": 45570
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.578148148148149e-05,
+      "loss": 0.0027,
+      "step": 45580
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.578055555555556e-05,
+      "loss": 0.0041,
+      "step": 45590
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 4.577962962962963e-05,
+      "loss": 0.0066,
+      "step": 45600
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 4.577870370370371e-05,
+      "loss": 0.0058,
+      "step": 45610
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 4.577777777777778e-05,
+      "loss": 0.0048,
+      "step": 45620
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 4.577685185185185e-05,
+      "loss": 0.0032,
+      "step": 45630
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 4.5775925925925926e-05,
+      "loss": 0.0038,
+      "step": 45640
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 4.5775e-05,
+      "loss": 0.0055,
+      "step": 45650
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.5774074074074076e-05,
+      "loss": 0.004,
+      "step": 45660
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.577314814814815e-05,
+      "loss": 0.0042,
+      "step": 45670
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.577222222222222e-05,
+      "loss": 0.0056,
+      "step": 45680
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.57712962962963e-05,
+      "loss": 0.0033,
+      "step": 45690
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.577037037037038e-05,
+      "loss": 0.0075,
+      "step": 45700
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.5769444444444445e-05,
+      "loss": 0.0056,
+      "step": 45710
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.576851851851852e-05,
+      "loss": 0.0054,
+      "step": 45720
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.5767592592592595e-05,
+      "loss": 0.0033,
+      "step": 45730
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.576666666666667e-05,
+      "loss": 0.0034,
+      "step": 45740
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.5765740740740745e-05,
+      "loss": 0.0042,
+      "step": 45750
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.5764814814814813e-05,
+      "loss": 0.0038,
+      "step": 45760
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.576388888888889e-05,
+      "loss": 0.0052,
+      "step": 45770
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.5762962962962964e-05,
+      "loss": 0.0039,
+      "step": 45780
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.576203703703704e-05,
+      "loss": 0.0042,
+      "step": 45790
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.5761111111111114e-05,
+      "loss": 0.0034,
+      "step": 45800
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.576018518518519e-05,
+      "loss": 0.003,
+      "step": 45810
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.5759259259259264e-05,
+      "loss": 0.0056,
+      "step": 45820
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.575833333333334e-05,
+      "loss": 0.0034,
+      "step": 45830
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.575740740740741e-05,
+      "loss": 0.0045,
+      "step": 45840
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.575648148148148e-05,
+      "loss": 0.0026,
+      "step": 45850
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.575555555555556e-05,
+      "loss": 0.0036,
+      "step": 45860
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.575462962962963e-05,
+      "loss": 0.0056,
+      "step": 45870
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.575370370370371e-05,
+      "loss": 0.0041,
+      "step": 45880
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.5752777777777776e-05,
+      "loss": 0.0035,
+      "step": 45890
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.575185185185185e-05,
+      "loss": 0.0063,
+      "step": 45900
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.575092592592593e-05,
+      "loss": 0.0049,
+      "step": 45910
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.575e-05,
+      "loss": 0.0048,
+      "step": 45920
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.5749074074074076e-05,
+      "loss": 0.0055,
+      "step": 45930
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.574814814814815e-05,
+      "loss": 0.0046,
+      "step": 45940
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.5747222222222226e-05,
+      "loss": 0.0041,
+      "step": 45950
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.57462962962963e-05,
+      "loss": 0.0036,
+      "step": 45960
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.574537037037037e-05,
+      "loss": 0.003,
+      "step": 45970
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.5744444444444444e-05,
+      "loss": 0.0048,
+      "step": 45980
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.574351851851852e-05,
+      "loss": 0.0039,
+      "step": 45990
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.5742592592592594e-05,
+      "loss": 0.0045,
+      "step": 46000
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.574166666666667e-05,
+      "loss": 0.0063,
+      "step": 46010
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.5740740740740745e-05,
+      "loss": 0.0044,
+      "step": 46020
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.573981481481482e-05,
+      "loss": 0.0035,
+      "step": 46030
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.5738888888888895e-05,
+      "loss": 0.0036,
+      "step": 46040
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.573796296296296e-05,
+      "loss": 0.0049,
+      "step": 46050
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.573703703703704e-05,
+      "loss": 0.0033,
+      "step": 46060
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.573611111111111e-05,
+      "loss": 0.0042,
+      "step": 46070
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.573518518518519e-05,
+      "loss": 0.0068,
+      "step": 46080
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.573425925925926e-05,
+      "loss": 0.0042,
+      "step": 46090
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.573333333333333e-05,
+      "loss": 0.0037,
+      "step": 46100
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.5732407407407406e-05,
+      "loss": 0.0063,
+      "step": 46110
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.573148148148149e-05,
+      "loss": 0.0034,
+      "step": 46120
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.5730555555555557e-05,
+      "loss": 0.0029,
+      "step": 46130
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.572962962962963e-05,
+      "loss": 0.0078,
+      "step": 46140
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.572870370370371e-05,
+      "loss": 0.0037,
+      "step": 46150
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.572777777777778e-05,
+      "loss": 0.0025,
+      "step": 46160
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.572685185185186e-05,
+      "loss": 0.0042,
+      "step": 46170
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.5725925925925925e-05,
+      "loss": 0.0039,
+      "step": 46180
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.5725e-05,
+      "loss": 0.0051,
+      "step": 46190
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.5724074074074075e-05,
+      "loss": 0.011,
+      "step": 46200
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.572314814814815e-05,
+      "loss": 0.0046,
+      "step": 46210
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.572222222222222e-05,
+      "loss": 0.0041,
+      "step": 46220
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.57212962962963e-05,
+      "loss": 0.0047,
+      "step": 46230
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.5720370370370375e-05,
+      "loss": 0.0079,
+      "step": 46240
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.571944444444445e-05,
+      "loss": 0.0056,
+      "step": 46250
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.571851851851852e-05,
+      "loss": 0.0029,
+      "step": 46260
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.5717592592592594e-05,
+      "loss": 0.0047,
+      "step": 46270
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.571666666666667e-05,
+      "loss": 0.006,
+      "step": 46280
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.5715740740740744e-05,
+      "loss": 0.003,
+      "step": 46290
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.571481481481482e-05,
+      "loss": 0.0036,
+      "step": 46300
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.571388888888889e-05,
+      "loss": 0.0055,
+      "step": 46310
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.571296296296296e-05,
+      "loss": 0.007,
+      "step": 46320
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.571203703703704e-05,
+      "loss": 0.0034,
+      "step": 46330
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.571111111111111e-05,
+      "loss": 0.0044,
+      "step": 46340
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.571018518518519e-05,
+      "loss": 0.0047,
+      "step": 46350
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.570925925925926e-05,
+      "loss": 0.0046,
+      "step": 46360
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.570833333333334e-05,
+      "loss": 0.003,
+      "step": 46370
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.570740740740741e-05,
+      "loss": 0.0072,
+      "step": 46380
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.570648148148148e-05,
+      "loss": 0.0037,
+      "step": 46390
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.5705555555555556e-05,
+      "loss": 0.0029,
+      "step": 46400
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.570462962962963e-05,
+      "loss": 0.0037,
+      "step": 46410
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.5703703703703706e-05,
+      "loss": 0.0031,
+      "step": 46420
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.5702777777777774e-05,
+      "loss": 0.0023,
+      "step": 46430
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.5701851851851856e-05,
+      "loss": 0.0033,
+      "step": 46440
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.570092592592593e-05,
+      "loss": 0.0071,
+      "step": 46450
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.5700000000000006e-05,
+      "loss": 0.0043,
+      "step": 46460
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.5699074074074075e-05,
+      "loss": 0.0028,
+      "step": 46470
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.569814814814815e-05,
+      "loss": 0.004,
+      "step": 46480
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.5697222222222225e-05,
+      "loss": 0.0034,
+      "step": 46490
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.56962962962963e-05,
+      "loss": 0.007,
+      "step": 46500
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.5695370370370375e-05,
+      "loss": 0.0035,
+      "step": 46510
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.569444444444444e-05,
+      "loss": 0.0048,
+      "step": 46520
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.569351851851852e-05,
+      "loss": 0.005,
+      "step": 46530
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.569259259259259e-05,
+      "loss": 0.0043,
+      "step": 46540
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.569166666666667e-05,
+      "loss": 0.0038,
+      "step": 46550
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.569074074074074e-05,
+      "loss": 0.0037,
+      "step": 46560
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.568981481481482e-05,
+      "loss": 0.0055,
+      "step": 46570
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.5688888888888893e-05,
+      "loss": 0.004,
+      "step": 46580
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.568796296296297e-05,
+      "loss": 0.0029,
+      "step": 46590
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.568703703703704e-05,
+      "loss": 0.004,
+      "step": 46600
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.568611111111111e-05,
+      "loss": 0.0031,
+      "step": 46610
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.568518518518519e-05,
+      "loss": 0.0052,
+      "step": 46620
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.568425925925926e-05,
+      "loss": 0.0048,
+      "step": 46630
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.568333333333333e-05,
+      "loss": 0.0046,
+      "step": 46640
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.5682407407407405e-05,
+      "loss": 0.0049,
+      "step": 46650
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.568148148148149e-05,
+      "loss": 0.0017,
+      "step": 46660
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.568055555555556e-05,
+      "loss": 0.0039,
+      "step": 46670
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.567962962962963e-05,
+      "loss": 0.0048,
+      "step": 46680
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.5678703703703705e-05,
+      "loss": 0.0047,
+      "step": 46690
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.567777777777778e-05,
+      "loss": 0.0045,
+      "step": 46700
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.5676851851851856e-05,
+      "loss": 0.0052,
+      "step": 46710
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.567592592592593e-05,
+      "loss": 0.004,
+      "step": 46720
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.5675e-05,
+      "loss": 0.0043,
+      "step": 46730
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.5674074074074074e-05,
+      "loss": 0.0041,
+      "step": 46740
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.567314814814815e-05,
+      "loss": 0.0045,
+      "step": 46750
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.5672222222222224e-05,
+      "loss": 0.0048,
+      "step": 46760
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.56712962962963e-05,
+      "loss": 0.0032,
+      "step": 46770
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.5670370370370374e-05,
+      "loss": 0.0037,
+      "step": 46780
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.566944444444445e-05,
+      "loss": 0.0039,
+      "step": 46790
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.5668518518518524e-05,
+      "loss": 0.0053,
+      "step": 46800
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.566759259259259e-05,
+      "loss": 0.0068,
+      "step": 46810
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.566666666666667e-05,
+      "loss": 0.0081,
+      "step": 46820
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.566574074074074e-05,
+      "loss": 0.0061,
+      "step": 46830
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.566481481481482e-05,
+      "loss": 0.0047,
+      "step": 46840
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.5663888888888886e-05,
+      "loss": 0.0085,
+      "step": 46850
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.566296296296296e-05,
+      "loss": 0.0074,
+      "step": 46860
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.5662037037037036e-05,
+      "loss": 0.0035,
+      "step": 46870
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.566111111111112e-05,
+      "loss": 0.0065,
+      "step": 46880
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.5660185185185186e-05,
+      "loss": 0.0051,
+      "step": 46890
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.565925925925926e-05,
+      "loss": 0.0047,
+      "step": 46900
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.5658333333333336e-05,
+      "loss": 0.0046,
+      "step": 46910
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.565740740740741e-05,
+      "loss": 0.0041,
+      "step": 46920
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.5656481481481486e-05,
+      "loss": 0.0033,
+      "step": 46930
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.5655555555555555e-05,
+      "loss": 0.0038,
+      "step": 46940
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.565462962962963e-05,
+      "loss": 0.006,
+      "step": 46950
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.5653703703703705e-05,
+      "loss": 0.0043,
+      "step": 46960
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.565277777777778e-05,
+      "loss": 0.0024,
+      "step": 46970
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.5651851851851855e-05,
+      "loss": 0.005,
+      "step": 46980
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.565092592592593e-05,
+      "loss": 0.0045,
+      "step": 46990
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.5650000000000005e-05,
+      "loss": 0.0041,
+      "step": 47000
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.564907407407408e-05,
+      "loss": 0.0057,
+      "step": 47010
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.564814814814815e-05,
+      "loss": 0.0026,
+      "step": 47020
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.5647222222222223e-05,
+      "loss": 0.0037,
+      "step": 47030
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.56462962962963e-05,
+      "loss": 0.0045,
+      "step": 47040
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.5645370370370374e-05,
+      "loss": 0.0057,
+      "step": 47050
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.564444444444444e-05,
+      "loss": 0.0037,
+      "step": 47060
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.564351851851852e-05,
+      "loss": 0.0053,
+      "step": 47070
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.564259259259259e-05,
+      "loss": 0.0036,
+      "step": 47080
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.5641666666666674e-05,
+      "loss": 0.0024,
+      "step": 47090
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.564074074074074e-05,
+      "loss": 0.0045,
+      "step": 47100
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.563981481481482e-05,
+      "loss": 0.0043,
+      "step": 47110
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.563888888888889e-05,
+      "loss": 0.0048,
+      "step": 47120
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.563796296296297e-05,
+      "loss": 0.0044,
+      "step": 47130
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.563703703703704e-05,
+      "loss": 0.0038,
+      "step": 47140
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.563611111111111e-05,
+      "loss": 0.0026,
+      "step": 47150
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.5635185185185186e-05,
+      "loss": 0.003,
+      "step": 47160
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.563425925925926e-05,
+      "loss": 0.0054,
+      "step": 47170
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.5633333333333336e-05,
+      "loss": 0.0028,
+      "step": 47180
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.5632407407407404e-05,
+      "loss": 0.0042,
+      "step": 47190
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.5631481481481486e-05,
+      "loss": 0.0029,
+      "step": 47200
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.563055555555556e-05,
+      "loss": 0.0035,
+      "step": 47210
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.5629629629629636e-05,
+      "loss": 0.0067,
+      "step": 47220
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.5628703703703704e-05,
+      "loss": 0.0054,
+      "step": 47230
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.562777777777778e-05,
+      "loss": 0.0031,
+      "step": 47240
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.5626851851851854e-05,
+      "loss": 0.0048,
+      "step": 47250
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.562592592592593e-05,
+      "loss": 0.0045,
+      "step": 47260
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.5625e-05,
+      "loss": 0.0049,
+      "step": 47270
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.562407407407407e-05,
+      "loss": 0.0024,
+      "step": 47280
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.562314814814815e-05,
+      "loss": 0.0043,
+      "step": 47290
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.562222222222222e-05,
+      "loss": 0.0051,
+      "step": 47300
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.56212962962963e-05,
+      "loss": 0.0056,
+      "step": 47310
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.562037037037037e-05,
+      "loss": 0.0046,
+      "step": 47320
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.561944444444445e-05,
+      "loss": 0.0059,
+      "step": 47330
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.561851851851852e-05,
+      "loss": 0.0058,
+      "step": 47340
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.56175925925926e-05,
+      "loss": 0.0023,
+      "step": 47350
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.5616666666666666e-05,
+      "loss": 0.0044,
+      "step": 47360
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.561574074074074e-05,
+      "loss": 0.0042,
+      "step": 47370
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.5614814814814817e-05,
+      "loss": 0.0037,
+      "step": 47380
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 4.561388888888889e-05,
+      "loss": 0.0053,
+      "step": 47390
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 4.561296296296296e-05,
+      "loss": 0.0043,
+      "step": 47400
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 4.5612037037037035e-05,
+      "loss": 0.0064,
+      "step": 47410
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 4.561111111111112e-05,
+      "loss": 0.0035,
+      "step": 47420
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 4.561018518518519e-05,
+      "loss": 0.0052,
+      "step": 47430
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 4.560925925925926e-05,
+      "loss": 0.0055,
+      "step": 47440
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 4.5608333333333335e-05,
+      "loss": 0.0037,
+      "step": 47450
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 4.560740740740741e-05,
+      "loss": 0.0029,
+      "step": 47460
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 4.5606481481481485e-05,
+      "loss": 0.0046,
+      "step": 47470
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 4.560555555555556e-05,
+      "loss": 0.0052,
+      "step": 47480
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 4.560462962962963e-05,
+      "loss": 0.0075,
+      "step": 47490
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 4.5603703703703704e-05,
+      "loss": 0.0034,
+      "step": 47500
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 4.560287037037037e-05,
+      "loss": 0.0059,
+      "step": 47510
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 4.5601944444444444e-05,
+      "loss": 0.0047,
+      "step": 47520
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 4.5601018518518525e-05,
+      "loss": 0.0069,
+      "step": 47530
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 4.5600092592592594e-05,
+      "loss": 0.0041,
+      "step": 47540
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 4.559916666666667e-05,
+      "loss": 0.0041,
+      "step": 47550
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 4.5598240740740744e-05,
+      "loss": 0.004,
+      "step": 47560
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 4.559731481481482e-05,
+      "loss": 0.0049,
+      "step": 47570
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 4.5596388888888894e-05,
+      "loss": 0.0022,
+      "step": 47580
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 4.559546296296296e-05,
+      "loss": 0.0042,
+      "step": 47590
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 4.559453703703704e-05,
+      "loss": 0.005,
+      "step": 47600
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 4.559361111111111e-05,
+      "loss": 0.0036,
+      "step": 47610
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 4.559268518518519e-05,
+      "loss": 0.0047,
+      "step": 47620
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 4.559175925925926e-05,
+      "loss": 0.0055,
+      "step": 47630
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 4.559083333333334e-05,
+      "loss": 0.0041,
+      "step": 47640
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 4.558990740740741e-05,
+      "loss": 0.0039,
+      "step": 47650
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 4.558898148148149e-05,
+      "loss": 0.0037,
+      "step": 47660
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 4.5588055555555556e-05,
+      "loss": 0.0046,
+      "step": 47670
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 4.558712962962963e-05,
+      "loss": 0.003,
+      "step": 47680
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 4.5586203703703706e-05,
+      "loss": 0.0048,
+      "step": 47690
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 4.558527777777778e-05,
+      "loss": 0.004,
+      "step": 47700
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 4.5584351851851856e-05,
+      "loss": 0.0068,
+      "step": 47710
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 4.5583425925925924e-05,
+      "loss": 0.0081,
+      "step": 47720
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 4.55825e-05,
+      "loss": 0.0042,
+      "step": 47730
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 4.5581574074074074e-05,
+      "loss": 0.0034,
+      "step": 47740
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 4.558064814814815e-05,
+      "loss": 0.0049,
+      "step": 47750
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 4.5579722222222225e-05,
+      "loss": 0.0037,
+      "step": 47760
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 4.55787962962963e-05,
+      "loss": 0.0042,
+      "step": 47770
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 4.5577870370370375e-05,
+      "loss": 0.0049,
+      "step": 47780
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 4.557694444444445e-05,
+      "loss": 0.0038,
+      "step": 47790
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 4.557601851851852e-05,
+      "loss": 0.0056,
+      "step": 47800
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 4.557509259259259e-05,
+      "loss": 0.0055,
+      "step": 47810
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 4.557416666666667e-05,
+      "loss": 0.0036,
+      "step": 47820
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 4.557324074074074e-05,
+      "loss": 0.0025,
+      "step": 47830
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 4.557231481481482e-05,
+      "loss": 0.0063,
+      "step": 47840
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 4.5571388888888886e-05,
+      "loss": 0.0063,
+      "step": 47850
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 4.557046296296297e-05,
+      "loss": 0.0037,
+      "step": 47860
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 4.556953703703704e-05,
+      "loss": 0.0067,
+      "step": 47870
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 4.556861111111111e-05,
+      "loss": 0.0022,
+      "step": 47880
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 4.556768518518519e-05,
+      "loss": 0.0039,
+      "step": 47890
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 4.556675925925926e-05,
+      "loss": 0.008,
+      "step": 47900
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 4.556583333333334e-05,
+      "loss": 0.0033,
+      "step": 47910
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 4.556490740740741e-05,
+      "loss": 0.0031,
+      "step": 47920
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 4.556398148148148e-05,
+      "loss": 0.0039,
+      "step": 47930
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 4.5563055555555555e-05,
+      "loss": 0.0045,
+      "step": 47940
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 4.556212962962963e-05,
+      "loss": 0.003,
+      "step": 47950
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 4.5561203703703705e-05,
+      "loss": 0.0039,
+      "step": 47960
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 4.556027777777778e-05,
+      "loss": 0.0044,
+      "step": 47970
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 4.5559351851851855e-05,
+      "loss": 0.0047,
+      "step": 47980
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 4.555842592592593e-05,
+      "loss": 0.0033,
+      "step": 47990
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 4.5557500000000006e-05,
+      "loss": 0.0068,
+      "step": 48000
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 4.5556574074074074e-05,
+      "loss": 0.0035,
+      "step": 48010
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 4.555564814814815e-05,
+      "loss": 0.0045,
+      "step": 48020
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 4.5554722222222224e-05,
+      "loss": 0.0054,
+      "step": 48030
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 4.55537962962963e-05,
+      "loss": 0.004,
+      "step": 48040
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 4.5552870370370374e-05,
+      "loss": 0.0042,
+      "step": 48050
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 4.555194444444444e-05,
+      "loss": 0.0058,
+      "step": 48060
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 4.5551018518518524e-05,
+      "loss": 0.0052,
+      "step": 48070
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 4.55500925925926e-05,
+      "loss": 0.0063,
+      "step": 48080
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 4.554916666666667e-05,
+      "loss": 0.0042,
+      "step": 48090
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 4.554824074074074e-05,
+      "loss": 0.0035,
+      "step": 48100
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 4.554731481481482e-05,
+      "loss": 0.0022,
+      "step": 48110
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 4.554638888888889e-05,
+      "loss": 0.0046,
+      "step": 48120
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 4.554546296296297e-05,
+      "loss": 0.0036,
+      "step": 48130
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 4.5544537037037036e-05,
+      "loss": 0.0018,
+      "step": 48140
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 4.554361111111111e-05,
+      "loss": 0.003,
+      "step": 48150
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 4.5542685185185186e-05,
+      "loss": 0.0071,
+      "step": 48160
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 4.554175925925926e-05,
+      "loss": 0.004,
+      "step": 48170
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 4.5540833333333336e-05,
+      "loss": 0.003,
+      "step": 48180
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 4.553990740740741e-05,
+      "loss": 0.005,
+      "step": 48190
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 4.5538981481481486e-05,
+      "loss": 0.0056,
+      "step": 48200
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 4.553805555555556e-05,
+      "loss": 0.0047,
+      "step": 48210
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 4.553712962962963e-05,
+      "loss": 0.0044,
+      "step": 48220
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 4.5536203703703705e-05,
+      "loss": 0.0026,
+      "step": 48230
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 4.553527777777778e-05,
+      "loss": 0.007,
+      "step": 48240
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 4.5534351851851855e-05,
+      "loss": 0.0033,
+      "step": 48250
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 4.553342592592593e-05,
+      "loss": 0.0052,
+      "step": 48260
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 4.55325e-05,
+      "loss": 0.0044,
+      "step": 48270
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 4.553157407407407e-05,
+      "loss": 0.003,
+      "step": 48280
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 4.5530648148148155e-05,
+      "loss": 0.0056,
+      "step": 48290
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 4.552972222222222e-05,
+      "loss": 0.0041,
+      "step": 48300
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 4.55287962962963e-05,
+      "loss": 0.0046,
+      "step": 48310
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 4.5527870370370373e-05,
+      "loss": 0.0039,
+      "step": 48320
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 4.552694444444445e-05,
+      "loss": 0.0055,
+      "step": 48330
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 4.5526018518518524e-05,
+      "loss": 0.0065,
+      "step": 48340
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 4.552509259259259e-05,
+      "loss": 0.0046,
+      "step": 48350
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 4.552416666666667e-05,
+      "loss": 0.0031,
+      "step": 48360
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 4.552324074074074e-05,
+      "loss": 0.0036,
+      "step": 48370
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 4.552231481481482e-05,
+      "loss": 0.0045,
+      "step": 48380
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 4.5521388888888885e-05,
+      "loss": 0.005,
+      "step": 48390
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 4.552046296296297e-05,
+      "loss": 0.005,
+      "step": 48400
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 4.551953703703704e-05,
+      "loss": 0.0049,
+      "step": 48410
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 4.551861111111112e-05,
+      "loss": 0.0059,
+      "step": 48420
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 4.5517685185185185e-05,
+      "loss": 0.001,
+      "step": 48430
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 4.551675925925926e-05,
+      "loss": 0.0058,
+      "step": 48440
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 4.5515833333333336e-05,
+      "loss": 0.0029,
+      "step": 48450
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 4.551490740740741e-05,
+      "loss": 0.0056,
+      "step": 48460
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 4.5513981481481486e-05,
+      "loss": 0.0038,
+      "step": 48470
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 4.5513055555555554e-05,
+      "loss": 0.0045,
+      "step": 48480
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 4.551212962962963e-05,
+      "loss": 0.0038,
+      "step": 48490
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 4.5511203703703704e-05,
+      "loss": 0.0097,
+      "step": 48500
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 4.551027777777778e-05,
+      "loss": 0.0057,
+      "step": 48510
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 4.5509351851851854e-05,
+      "loss": 0.0053,
+      "step": 48520
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 4.550842592592593e-05,
+      "loss": 0.0051,
+      "step": 48530
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 4.5507500000000004e-05,
+      "loss": 0.0041,
+      "step": 48540
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 4.550657407407408e-05,
+      "loss": 0.003,
+      "step": 48550
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 4.550564814814815e-05,
+      "loss": 0.0047,
+      "step": 48560
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 4.550472222222222e-05,
+      "loss": 0.0027,
+      "step": 48570
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 4.55037962962963e-05,
+      "loss": 0.0057,
+      "step": 48580
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 4.550287037037037e-05,
+      "loss": 0.0042,
+      "step": 48590
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 4.550194444444444e-05,
+      "loss": 0.0035,
+      "step": 48600
+    },
+    {
+      "epoch": 9.0,
+      "eval_loss": 0.014685405418276787,
+      "eval_runtime": 154.2291,
+      "eval_samples_per_second": 25.871,
+      "eval_steps_per_second": 3.235,
+      "step": 48600
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 4.550101851851852e-05,
+      "loss": 0.0054,
+      "step": 48610
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 4.55000925925926e-05,
+      "loss": 0.0023,
+      "step": 48620
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 4.549916666666667e-05,
+      "loss": 0.0044,
+      "step": 48630
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 4.549824074074074e-05,
+      "loss": 0.0034,
+      "step": 48640
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 4.5497314814814816e-05,
+      "loss": 0.0038,
+      "step": 48650
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 4.549638888888889e-05,
+      "loss": 0.003,
+      "step": 48660
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 4.5495462962962966e-05,
+      "loss": 0.0033,
+      "step": 48670
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 4.549453703703704e-05,
+      "loss": 0.0039,
+      "step": 48680
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 4.549361111111111e-05,
+      "loss": 0.0025,
+      "step": 48690
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 4.5492685185185185e-05,
+      "loss": 0.0046,
+      "step": 48700
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 4.549175925925926e-05,
+      "loss": 0.0033,
+      "step": 48710
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 4.5490833333333335e-05,
+      "loss": 0.0031,
+      "step": 48720
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 4.548990740740741e-05,
+      "loss": 0.0031,
+      "step": 48730
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 4.5488981481481485e-05,
+      "loss": 0.0027,
+      "step": 48740
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 4.548805555555556e-05,
+      "loss": 0.0053,
+      "step": 48750
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 4.5487129629629635e-05,
+      "loss": 0.0042,
+      "step": 48760
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 4.5486203703703703e-05,
+      "loss": 0.0027,
+      "step": 48770
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 4.548527777777778e-05,
+      "loss": 0.0025,
+      "step": 48780
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.5484351851851854e-05,
+      "loss": 0.0046,
+      "step": 48790
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.548342592592593e-05,
+      "loss": 0.0058,
+      "step": 48800
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.54825e-05,
+      "loss": 0.002,
+      "step": 48810
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.548157407407407e-05,
+      "loss": 0.0035,
+      "step": 48820
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.5480648148148154e-05,
+      "loss": 0.0055,
+      "step": 48830
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.547972222222223e-05,
+      "loss": 0.0049,
+      "step": 48840
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.54787962962963e-05,
+      "loss": 0.0029,
+      "step": 48850
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.547787037037037e-05,
+      "loss": 0.0031,
+      "step": 48860
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.547694444444445e-05,
+      "loss": 0.0019,
+      "step": 48870
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.547601851851852e-05,
+      "loss": 0.0089,
+      "step": 48880
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.54750925925926e-05,
+      "loss": 0.0039,
+      "step": 48890
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.5474166666666666e-05,
+      "loss": 0.0026,
+      "step": 48900
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.547324074074074e-05,
+      "loss": 0.0027,
+      "step": 48910
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.5472314814814816e-05,
+      "loss": 0.0036,
+      "step": 48920
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.547138888888889e-05,
+      "loss": 0.0038,
+      "step": 48930
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.5470462962962966e-05,
+      "loss": 0.0034,
+      "step": 48940
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.546953703703704e-05,
+      "loss": 0.0043,
+      "step": 48950
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.5468611111111116e-05,
+      "loss": 0.0016,
+      "step": 48960
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.546768518518519e-05,
+      "loss": 0.003,
+      "step": 48970
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.546675925925926e-05,
+      "loss": 0.0037,
+      "step": 48980
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.5465833333333334e-05,
+      "loss": 0.0026,
+      "step": 48990
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.546490740740741e-05,
+      "loss": 0.0042,
+      "step": 49000
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 4.5463981481481485e-05,
+      "loss": 0.0022,
+      "step": 49010
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 4.546305555555555e-05,
+      "loss": 0.0039,
+      "step": 49020
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 4.546212962962963e-05,
+      "loss": 0.0053,
+      "step": 49030
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 4.546120370370371e-05,
+      "loss": 0.0031,
+      "step": 49040
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 4.5460277777777785e-05,
+      "loss": 0.0068,
+      "step": 49050
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.545935185185185e-05,
+      "loss": 0.0054,
+      "step": 49060
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.545842592592593e-05,
+      "loss": 0.0041,
+      "step": 49070
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.54575e-05,
+      "loss": 0.0027,
+      "step": 49080
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.545657407407408e-05,
+      "loss": 0.0024,
+      "step": 49090
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.545564814814815e-05,
+      "loss": 0.0035,
+      "step": 49100
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.545472222222222e-05,
+      "loss": 0.0049,
+      "step": 49110
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.5453796296296297e-05,
+      "loss": 0.0026,
+      "step": 49120
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.545287037037037e-05,
+      "loss": 0.0066,
+      "step": 49130
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.545194444444445e-05,
+      "loss": 0.0054,
+      "step": 49140
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.545101851851852e-05,
+      "loss": 0.0043,
+      "step": 49150
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.54500925925926e-05,
+      "loss": 0.0025,
+      "step": 49160
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 4.544916666666667e-05,
+      "loss": 0.0034,
+      "step": 49170
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 4.544824074074075e-05,
+      "loss": 0.0033,
+      "step": 49180
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 4.5447314814814815e-05,
+      "loss": 0.0058,
+      "step": 49190
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 4.544638888888889e-05,
+      "loss": 0.0049,
+      "step": 49200
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 4.5445462962962965e-05,
+      "loss": 0.003,
+      "step": 49210
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 4.544453703703704e-05,
+      "loss": 0.0035,
+      "step": 49220
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.544361111111111e-05,
+      "loss": 0.004,
+      "step": 49230
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.5442685185185184e-05,
+      "loss": 0.0042,
+      "step": 49240
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.544175925925926e-05,
+      "loss": 0.0032,
+      "step": 49250
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.544083333333334e-05,
+      "loss": 0.0038,
+      "step": 49260
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.543990740740741e-05,
+      "loss": 0.0043,
+      "step": 49270
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 4.5438981481481484e-05,
+      "loss": 0.0033,
+      "step": 49280
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 4.543805555555556e-05,
+      "loss": 0.0037,
+      "step": 49290
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 4.5437129629629634e-05,
+      "loss": 0.0025,
+      "step": 49300
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 4.543620370370371e-05,
+      "loss": 0.0041,
+      "step": 49310
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 4.543527777777778e-05,
+      "loss": 0.0036,
+      "step": 49320
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 4.543435185185185e-05,
+      "loss": 0.0019,
+      "step": 49330
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 4.543342592592593e-05,
+      "loss": 0.0035,
+      "step": 49340
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 4.54325e-05,
+      "loss": 0.0038,
+      "step": 49350
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 4.543157407407407e-05,
+      "loss": 0.0067,
+      "step": 49360
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 4.543064814814815e-05,
+      "loss": 0.0052,
+      "step": 49370
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 4.542972222222223e-05,
+      "loss": 0.0031,
+      "step": 49380
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 4.54287962962963e-05,
+      "loss": 0.0039,
+      "step": 49390
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 4.542787037037037e-05,
+      "loss": 0.0042,
+      "step": 49400
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 4.5426944444444446e-05,
+      "loss": 0.0055,
+      "step": 49410
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 4.542601851851852e-05,
+      "loss": 0.0051,
+      "step": 49420
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 4.5425092592592596e-05,
+      "loss": 0.0043,
+      "step": 49430
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 4.5424166666666664e-05,
+      "loss": 0.0037,
+      "step": 49440
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 4.542324074074074e-05,
+      "loss": 0.0047,
+      "step": 49450
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 4.5422314814814815e-05,
+      "loss": 0.004,
+      "step": 49460
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 4.542138888888889e-05,
+      "loss": 0.0033,
+      "step": 49470
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 4.5420462962962965e-05,
+      "loss": 0.0032,
+      "step": 49480
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 4.541953703703704e-05,
+      "loss": 0.0052,
+      "step": 49490
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 4.5418611111111115e-05,
+      "loss": 0.0048,
+      "step": 49500
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 4.541768518518519e-05,
+      "loss": 0.0032,
+      "step": 49510
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 4.5416759259259265e-05,
+      "loss": 0.0029,
+      "step": 49520
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 4.541583333333333e-05,
+      "loss": 0.0056,
+      "step": 49530
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 4.541490740740741e-05,
+      "loss": 0.0027,
+      "step": 49540
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 4.541398148148148e-05,
+      "loss": 0.0032,
+      "step": 49550
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 4.541305555555556e-05,
+      "loss": 0.0018,
+      "step": 49560
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 4.541212962962963e-05,
+      "loss": 0.0049,
+      "step": 49570
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 4.541120370370371e-05,
+      "loss": 0.003,
+      "step": 49580
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 4.5410277777777784e-05,
+      "loss": 0.0039,
+      "step": 49590
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 4.540935185185186e-05,
+      "loss": 0.0022,
+      "step": 49600
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 4.540842592592593e-05,
+      "loss": 0.0051,
+      "step": 49610
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 4.54075e-05,
+      "loss": 0.0027,
+      "step": 49620
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 4.540657407407408e-05,
+      "loss": 0.0048,
+      "step": 49630
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 4.540564814814815e-05,
+      "loss": 0.003,
+      "step": 49640
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 4.540472222222222e-05,
+      "loss": 0.0045,
+      "step": 49650
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 4.5403796296296295e-05,
+      "loss": 0.0025,
+      "step": 49660
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 4.540287037037037e-05,
+      "loss": 0.0031,
+      "step": 49670
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 4.5401944444444445e-05,
+      "loss": 0.0018,
+      "step": 49680
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 4.540101851851852e-05,
+      "loss": 0.005,
+      "step": 49690
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 4.5400092592592596e-05,
+      "loss": 0.0047,
+      "step": 49700
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 4.539916666666667e-05,
+      "loss": 0.0045,
+      "step": 49710
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 4.5398240740740746e-05,
+      "loss": 0.0034,
+      "step": 49720
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 4.539731481481482e-05,
+      "loss": 0.0081,
+      "step": 49730
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 4.539638888888889e-05,
+      "loss": 0.0025,
+      "step": 49740
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 4.5395462962962964e-05,
+      "loss": 0.004,
+      "step": 49750
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 4.539453703703704e-05,
+      "loss": 0.0048,
+      "step": 49760
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 4.5393611111111114e-05,
+      "loss": 0.0057,
+      "step": 49770
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 4.539268518518518e-05,
+      "loss": 0.0025,
+      "step": 49780
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 4.539175925925926e-05,
+      "loss": 0.0043,
+      "step": 49790
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 4.539083333333334e-05,
+      "loss": 0.0044,
+      "step": 49800
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 4.5389907407407414e-05,
+      "loss": 0.0023,
+      "step": 49810
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 4.538898148148148e-05,
+      "loss": 0.0036,
+      "step": 49820
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 4.538805555555556e-05,
+      "loss": 0.0022,
+      "step": 49830
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 4.538712962962963e-05,
+      "loss": 0.0045,
+      "step": 49840
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 4.538620370370371e-05,
+      "loss": 0.0026,
+      "step": 49850
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 4.5385277777777776e-05,
+      "loss": 0.002,
+      "step": 49860
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 4.538435185185185e-05,
+      "loss": 0.0024,
+      "step": 49870
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 4.5383425925925926e-05,
+      "loss": 0.0038,
+      "step": 49880
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 4.53825e-05,
+      "loss": 0.0015,
+      "step": 49890
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 4.5381574074074076e-05,
+      "loss": 0.0052,
+      "step": 49900
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 4.538064814814815e-05,
+      "loss": 0.0035,
+      "step": 49910
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 4.5379722222222226e-05,
+      "loss": 0.0055,
+      "step": 49920
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 4.53787962962963e-05,
+      "loss": 0.006,
+      "step": 49930
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 4.5377870370370377e-05,
+      "loss": 0.004,
+      "step": 49940
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 4.5376944444444445e-05,
+      "loss": 0.0066,
+      "step": 49950
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 4.537601851851852e-05,
+      "loss": 0.0055,
+      "step": 49960
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 4.5375092592592595e-05,
+      "loss": 0.006,
+      "step": 49970
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 4.537416666666667e-05,
+      "loss": 0.0032,
+      "step": 49980
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 4.537324074074074e-05,
+      "loss": 0.0038,
+      "step": 49990
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 4.537231481481481e-05,
+      "loss": 0.0031,
+      "step": 50000
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 4.5371388888888895e-05,
+      "loss": 0.0035,
+      "step": 50010
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 4.537046296296297e-05,
+      "loss": 0.0069,
+      "step": 50020
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 4.536953703703704e-05,
+      "loss": 0.0035,
+      "step": 50030
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 4.5368611111111114e-05,
+      "loss": 0.0054,
+      "step": 50040
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 4.536768518518519e-05,
+      "loss": 0.004,
+      "step": 50050
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 4.5366759259259264e-05,
+      "loss": 0.0058,
+      "step": 50060
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 4.536583333333333e-05,
+      "loss": 0.0038,
+      "step": 50070
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 4.536490740740741e-05,
+      "loss": 0.0045,
+      "step": 50080
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 4.536398148148148e-05,
+      "loss": 0.0028,
+      "step": 50090
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 4.536305555555556e-05,
+      "loss": 0.004,
+      "step": 50100
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 4.536212962962963e-05,
+      "loss": 0.0027,
+      "step": 50110
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 4.536120370370371e-05,
+      "loss": 0.0023,
+      "step": 50120
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 4.536027777777778e-05,
+      "loss": 0.0037,
+      "step": 50130
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 4.535935185185186e-05,
+      "loss": 0.0044,
+      "step": 50140
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 4.535842592592593e-05,
+      "loss": 0.0041,
+      "step": 50150
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 4.53575e-05,
+      "loss": 0.0038,
+      "step": 50160
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 4.5356574074074076e-05,
+      "loss": 0.0023,
+      "step": 50170
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 4.535564814814815e-05,
+      "loss": 0.007,
+      "step": 50180
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 4.5354722222222226e-05,
+      "loss": 0.0034,
+      "step": 50190
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 4.5353796296296294e-05,
+      "loss": 0.0043,
+      "step": 50200
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 4.535287037037037e-05,
+      "loss": 0.0057,
+      "step": 50210
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 4.5351944444444444e-05,
+      "loss": 0.002,
+      "step": 50220
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 4.5351018518518526e-05,
+      "loss": 0.0059,
+      "step": 50230
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 4.5350092592592594e-05,
+      "loss": 0.0029,
+      "step": 50240
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 4.534916666666667e-05,
+      "loss": 0.0077,
+      "step": 50250
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 4.5348240740740744e-05,
+      "loss": 0.0028,
+      "step": 50260
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 4.534731481481482e-05,
+      "loss": 0.0041,
+      "step": 50270
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 4.534638888888889e-05,
+      "loss": 0.0029,
+      "step": 50280
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 4.534546296296296e-05,
+      "loss": 0.0038,
+      "step": 50290
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 4.534453703703704e-05,
+      "loss": 0.0046,
+      "step": 50300
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 4.534361111111111e-05,
+      "loss": 0.0039,
+      "step": 50310
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 4.534268518518519e-05,
+      "loss": 0.0042,
+      "step": 50320
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 4.5341759259259256e-05,
+      "loss": 0.0056,
+      "step": 50330
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 4.534083333333334e-05,
+      "loss": 0.0062,
+      "step": 50340
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 4.533990740740741e-05,
+      "loss": 0.0031,
+      "step": 50350
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 4.533898148148149e-05,
+      "loss": 0.003,
+      "step": 50360
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 4.5338055555555557e-05,
+      "loss": 0.0026,
+      "step": 50370
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 4.533712962962963e-05,
+      "loss": 0.0034,
+      "step": 50380
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 4.533620370370371e-05,
+      "loss": 0.003,
+      "step": 50390
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 4.533527777777778e-05,
+      "loss": 0.0036,
+      "step": 50400
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 4.533435185185185e-05,
+      "loss": 0.0034,
+      "step": 50410
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 4.5333425925925925e-05,
+      "loss": 0.0029,
+      "step": 50420
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 4.53325e-05,
+      "loss": 0.003,
+      "step": 50430
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 4.5331574074074075e-05,
+      "loss": 0.0041,
+      "step": 50440
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 4.533064814814815e-05,
+      "loss": 0.004,
+      "step": 50450
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 4.5329722222222225e-05,
+      "loss": 0.0055,
+      "step": 50460
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 4.53287962962963e-05,
+      "loss": 0.005,
+      "step": 50470
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 4.5327870370370375e-05,
+      "loss": 0.0044,
+      "step": 50480
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 4.5326944444444444e-05,
+      "loss": 0.0027,
+      "step": 50490
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 4.532601851851852e-05,
+      "loss": 0.003,
+      "step": 50500
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 4.5325092592592594e-05,
+      "loss": 0.0072,
+      "step": 50510
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 4.532416666666667e-05,
+      "loss": 0.006,
+      "step": 50520
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 4.5323240740740744e-05,
+      "loss": 0.0025,
+      "step": 50530
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 4.532231481481481e-05,
+      "loss": 0.0055,
+      "step": 50540
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 4.5321388888888894e-05,
+      "loss": 0.004,
+      "step": 50550
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 4.532046296296297e-05,
+      "loss": 0.0059,
+      "step": 50560
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 4.5319537037037044e-05,
+      "loss": 0.0024,
+      "step": 50570
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 4.531861111111111e-05,
+      "loss": 0.0034,
+      "step": 50580
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 4.531768518518519e-05,
+      "loss": 0.0069,
+      "step": 50590
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 4.531675925925926e-05,
+      "loss": 0.0043,
+      "step": 50600
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 4.531583333333334e-05,
+      "loss": 0.0041,
+      "step": 50610
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 4.5314907407407406e-05,
+      "loss": 0.0056,
+      "step": 50620
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 4.531398148148148e-05,
+      "loss": 0.0048,
+      "step": 50630
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 4.5313055555555556e-05,
+      "loss": 0.0033,
+      "step": 50640
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 4.531212962962963e-05,
+      "loss": 0.0054,
+      "step": 50650
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 4.5311203703703706e-05,
+      "loss": 0.0056,
+      "step": 50660
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 4.531027777777778e-05,
+      "loss": 0.0039,
+      "step": 50670
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 4.5309351851851856e-05,
+      "loss": 0.0029,
+      "step": 50680
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 4.530842592592593e-05,
+      "loss": 0.0066,
+      "step": 50690
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 4.53075e-05,
+      "loss": 0.0035,
+      "step": 50700
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 4.5306574074074075e-05,
+      "loss": 0.0038,
+      "step": 50710
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 4.530564814814815e-05,
+      "loss": 0.0024,
+      "step": 50720
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 4.5304722222222225e-05,
+      "loss": 0.0057,
+      "step": 50730
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 4.53037962962963e-05,
+      "loss": 0.0019,
+      "step": 50740
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 4.530287037037037e-05,
+      "loss": 0.0056,
+      "step": 50750
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 4.530194444444444e-05,
+      "loss": 0.0035,
+      "step": 50760
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 4.5301018518518525e-05,
+      "loss": 0.0055,
+      "step": 50770
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 4.53000925925926e-05,
+      "loss": 0.0035,
+      "step": 50780
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 4.529916666666667e-05,
+      "loss": 0.0029,
+      "step": 50790
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 4.529824074074074e-05,
+      "loss": 0.0039,
+      "step": 50800
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 4.529731481481482e-05,
+      "loss": 0.0039,
+      "step": 50810
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 4.529638888888889e-05,
+      "loss": 0.0044,
+      "step": 50820
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 4.529546296296296e-05,
+      "loss": 0.0074,
+      "step": 50830
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 4.529453703703704e-05,
+      "loss": 0.0046,
+      "step": 50840
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 4.529361111111111e-05,
+      "loss": 0.0036,
+      "step": 50850
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 4.529268518518519e-05,
+      "loss": 0.0037,
+      "step": 50860
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 4.529175925925926e-05,
+      "loss": 0.0054,
+      "step": 50870
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 4.529083333333334e-05,
+      "loss": 0.003,
+      "step": 50880
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 4.528990740740741e-05,
+      "loss": 0.0033,
+      "step": 50890
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 4.528898148148149e-05,
+      "loss": 0.0039,
+      "step": 50900
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 4.5288055555555555e-05,
+      "loss": 0.0029,
+      "step": 50910
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 4.528712962962963e-05,
+      "loss": 0.0059,
+      "step": 50920
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 4.5286203703703705e-05,
+      "loss": 0.0034,
+      "step": 50930
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 4.528527777777778e-05,
+      "loss": 0.0037,
+      "step": 50940
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 4.5284351851851856e-05,
+      "loss": 0.0027,
+      "step": 50950
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 4.5283425925925924e-05,
+      "loss": 0.0087,
+      "step": 50960
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 4.52825e-05,
+      "loss": 0.0027,
+      "step": 50970
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 4.5281574074074074e-05,
+      "loss": 0.006,
+      "step": 50980
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 4.5280648148148156e-05,
+      "loss": 0.004,
+      "step": 50990
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 4.5279722222222224e-05,
+      "loss": 0.0064,
+      "step": 51000
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 4.52787962962963e-05,
+      "loss": 0.0043,
+      "step": 51010
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 4.5277870370370374e-05,
+      "loss": 0.0081,
+      "step": 51020
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 4.527694444444445e-05,
+      "loss": 0.0053,
+      "step": 51030
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 4.527601851851852e-05,
+      "loss": 0.004,
+      "step": 51040
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 4.527509259259259e-05,
+      "loss": 0.0022,
+      "step": 51050
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 4.527416666666667e-05,
+      "loss": 0.0051,
+      "step": 51060
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 4.527324074074074e-05,
+      "loss": 0.0047,
+      "step": 51070
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 4.527231481481482e-05,
+      "loss": 0.0046,
+      "step": 51080
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 4.527138888888889e-05,
+      "loss": 0.0054,
+      "step": 51090
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 4.527046296296297e-05,
+      "loss": 0.0036,
+      "step": 51100
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 4.526953703703704e-05,
+      "loss": 0.0049,
+      "step": 51110
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 4.526861111111111e-05,
+      "loss": 0.0068,
+      "step": 51120
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 4.5267685185185186e-05,
+      "loss": 0.0034,
+      "step": 51130
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 4.526675925925926e-05,
+      "loss": 0.0036,
+      "step": 51140
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 4.5265833333333336e-05,
+      "loss": 0.004,
+      "step": 51150
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 4.526490740740741e-05,
+      "loss": 0.0054,
+      "step": 51160
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 4.526398148148148e-05,
+      "loss": 0.0034,
+      "step": 51170
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 4.5263055555555555e-05,
+      "loss": 0.0048,
+      "step": 51180
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 4.526212962962963e-05,
+      "loss": 0.0024,
+      "step": 51190
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 4.526120370370371e-05,
+      "loss": 0.003,
+      "step": 51200
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 4.526027777777778e-05,
+      "loss": 0.0037,
+      "step": 51210
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 4.5259351851851855e-05,
+      "loss": 0.0028,
+      "step": 51220
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 4.525842592592593e-05,
+      "loss": 0.0035,
+      "step": 51230
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 4.5257500000000005e-05,
+      "loss": 0.0042,
+      "step": 51240
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 4.525657407407407e-05,
+      "loss": 0.0044,
+      "step": 51250
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 4.525564814814815e-05,
+      "loss": 0.0029,
+      "step": 51260
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 4.5254722222222223e-05,
+      "loss": 0.0035,
+      "step": 51270
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 4.52537962962963e-05,
+      "loss": 0.0038,
+      "step": 51280
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 4.5252870370370374e-05,
+      "loss": 0.0057,
+      "step": 51290
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 4.525194444444444e-05,
+      "loss": 0.0069,
+      "step": 51300
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 4.5251018518518524e-05,
+      "loss": 0.0035,
+      "step": 51310
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 4.52500925925926e-05,
+      "loss": 0.0019,
+      "step": 51320
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 4.524916666666667e-05,
+      "loss": 0.0039,
+      "step": 51330
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 4.524824074074074e-05,
+      "loss": 0.0024,
+      "step": 51340
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 4.524731481481482e-05,
+      "loss": 0.0057,
+      "step": 51350
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 4.524638888888889e-05,
+      "loss": 0.0033,
+      "step": 51360
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 4.524546296296297e-05,
+      "loss": 0.0055,
+      "step": 51370
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 4.5244537037037035e-05,
+      "loss": 0.0033,
+      "step": 51380
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 4.524361111111111e-05,
+      "loss": 0.0039,
+      "step": 51390
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 4.5242685185185186e-05,
+      "loss": 0.0067,
+      "step": 51400
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 4.524175925925926e-05,
+      "loss": 0.0049,
+      "step": 51410
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 4.5240833333333336e-05,
+      "loss": 0.0054,
+      "step": 51420
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 4.523990740740741e-05,
+      "loss": 0.0052,
+      "step": 51430
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 4.5238981481481486e-05,
+      "loss": 0.0096,
+      "step": 51440
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 4.523805555555556e-05,
+      "loss": 0.0053,
+      "step": 51450
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 4.523712962962963e-05,
+      "loss": 0.0028,
+      "step": 51460
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 4.5236203703703704e-05,
+      "loss": 0.0042,
+      "step": 51470
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 4.523527777777778e-05,
+      "loss": 0.0042,
+      "step": 51480
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 4.5234351851851854e-05,
+      "loss": 0.004,
+      "step": 51490
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 4.523342592592593e-05,
+      "loss": 0.005,
+      "step": 51500
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 4.52325e-05,
+      "loss": 0.0048,
+      "step": 51510
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 4.523157407407408e-05,
+      "loss": 0.0028,
+      "step": 51520
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 4.5230648148148155e-05,
+      "loss": 0.0025,
+      "step": 51530
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 4.522972222222222e-05,
+      "loss": 0.0051,
+      "step": 51540
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 4.52287962962963e-05,
+      "loss": 0.0031,
+      "step": 51550
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 4.522787037037037e-05,
+      "loss": 0.0038,
+      "step": 51560
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 4.522694444444445e-05,
+      "loss": 0.007,
+      "step": 51570
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 4.522601851851852e-05,
+      "loss": 0.0038,
+      "step": 51580
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 4.522509259259259e-05,
+      "loss": 0.0039,
+      "step": 51590
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 4.5224166666666666e-05,
+      "loss": 0.0024,
+      "step": 51600
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 4.522324074074074e-05,
+      "loss": 0.0063,
+      "step": 51610
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 4.5222314814814816e-05,
+      "loss": 0.0029,
+      "step": 51620
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 4.522138888888889e-05,
+      "loss": 0.0046,
+      "step": 51630
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 4.5220462962962967e-05,
+      "loss": 0.0045,
+      "step": 51640
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 4.521953703703704e-05,
+      "loss": 0.0042,
+      "step": 51650
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 4.521861111111112e-05,
+      "loss": 0.0047,
+      "step": 51660
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 4.5217685185185185e-05,
+      "loss": 0.0038,
+      "step": 51670
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 4.521675925925926e-05,
+      "loss": 0.0024,
+      "step": 51680
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 4.5215833333333335e-05,
+      "loss": 0.0038,
+      "step": 51690
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 4.521490740740741e-05,
+      "loss": 0.0032,
+      "step": 51700
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 4.5213981481481485e-05,
+      "loss": 0.0057,
+      "step": 51710
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 4.5213055555555553e-05,
+      "loss": 0.0051,
+      "step": 51720
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 4.521212962962963e-05,
+      "loss": 0.0055,
+      "step": 51730
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 4.521120370370371e-05,
+      "loss": 0.0065,
+      "step": 51740
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 4.521027777777778e-05,
+      "loss": 0.0029,
+      "step": 51750
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 4.5209351851851854e-05,
+      "loss": 0.0035,
+      "step": 51760
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 4.520842592592593e-05,
+      "loss": 0.0026,
+      "step": 51770
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 4.5207500000000004e-05,
+      "loss": 0.0028,
+      "step": 51780
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 4.520657407407408e-05,
+      "loss": 0.0032,
+      "step": 51790
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 4.520564814814815e-05,
+      "loss": 0.0028,
+      "step": 51800
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 4.520472222222222e-05,
+      "loss": 0.0059,
+      "step": 51810
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 4.52037962962963e-05,
+      "loss": 0.0024,
+      "step": 51820
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 4.520287037037037e-05,
+      "loss": 0.0033,
+      "step": 51830
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 4.520194444444445e-05,
+      "loss": 0.0037,
+      "step": 51840
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 4.520101851851852e-05,
+      "loss": 0.0019,
+      "step": 51850
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 4.52000925925926e-05,
+      "loss": 0.0043,
+      "step": 51860
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 4.519916666666667e-05,
+      "loss": 0.0036,
+      "step": 51870
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 4.519824074074074e-05,
+      "loss": 0.0083,
+      "step": 51880
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 4.5197314814814816e-05,
+      "loss": 0.0035,
+      "step": 51890
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 4.519638888888889e-05,
+      "loss": 0.0034,
+      "step": 51900
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 4.5195462962962966e-05,
+      "loss": 0.0037,
+      "step": 51910
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 4.519453703703704e-05,
+      "loss": 0.0034,
+      "step": 51920
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 4.519361111111111e-05,
+      "loss": 0.0039,
+      "step": 51930
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 4.5192685185185184e-05,
+      "loss": 0.0046,
+      "step": 51940
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 4.519175925925926e-05,
+      "loss": 0.0034,
+      "step": 51950
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 4.5190833333333334e-05,
+      "loss": 0.0033,
+      "step": 51960
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 4.518990740740741e-05,
+      "loss": 0.005,
+      "step": 51970
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 4.5188981481481485e-05,
+      "loss": 0.0046,
+      "step": 51980
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 4.518805555555556e-05,
+      "loss": 0.0039,
+      "step": 51990
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 4.5187129629629635e-05,
+      "loss": 0.0042,
+      "step": 52000
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 4.51862037037037e-05,
+      "loss": 0.0052,
+      "step": 52010
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 4.518527777777778e-05,
+      "loss": 0.0053,
+      "step": 52020
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 4.518435185185185e-05,
+      "loss": 0.0043,
+      "step": 52030
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 4.518342592592593e-05,
+      "loss": 0.0022,
+      "step": 52040
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 4.51825e-05,
+      "loss": 0.0037,
+      "step": 52050
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 4.518157407407408e-05,
+      "loss": 0.0035,
+      "step": 52060
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 4.518064814814815e-05,
+      "loss": 0.003,
+      "step": 52070
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 4.517972222222223e-05,
+      "loss": 0.0076,
+      "step": 52080
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 4.51787962962963e-05,
+      "loss": 0.0046,
+      "step": 52090
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 4.517787037037037e-05,
+      "loss": 0.0031,
+      "step": 52100
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 4.517694444444445e-05,
+      "loss": 0.0041,
+      "step": 52110
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 4.517601851851852e-05,
+      "loss": 0.0046,
+      "step": 52120
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 4.51750925925926e-05,
+      "loss": 0.0055,
+      "step": 52130
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 4.5174166666666665e-05,
+      "loss": 0.0053,
+      "step": 52140
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 4.517324074074074e-05,
+      "loss": 0.0048,
+      "step": 52150
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 4.5172314814814815e-05,
+      "loss": 0.0027,
+      "step": 52160
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 4.517138888888889e-05,
+      "loss": 0.0033,
+      "step": 52170
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 4.5170462962962965e-05,
+      "loss": 0.0045,
+      "step": 52180
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 4.516953703703704e-05,
+      "loss": 0.0096,
+      "step": 52190
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 4.5168611111111115e-05,
+      "loss": 0.0033,
+      "step": 52200
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 4.516768518518519e-05,
+      "loss": 0.0055,
+      "step": 52210
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 4.516675925925926e-05,
+      "loss": 0.006,
+      "step": 52220
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 4.5165833333333334e-05,
+      "loss": 0.0033,
+      "step": 52230
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 4.516490740740741e-05,
+      "loss": 0.0037,
+      "step": 52240
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 4.5163981481481484e-05,
+      "loss": 0.0075,
+      "step": 52250
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 4.516305555555556e-05,
+      "loss": 0.0056,
+      "step": 52260
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 4.516212962962963e-05,
+      "loss": 0.0047,
+      "step": 52270
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 4.516120370370371e-05,
+      "loss": 0.0087,
+      "step": 52280
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 4.5160277777777784e-05,
+      "loss": 0.007,
+      "step": 52290
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 4.515944444444445e-05,
+      "loss": 0.0033,
+      "step": 52300
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 4.5158518518518524e-05,
+      "loss": 0.004,
+      "step": 52310
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 4.51575925925926e-05,
+      "loss": 0.0046,
+      "step": 52320
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 4.515666666666667e-05,
+      "loss": 0.004,
+      "step": 52330
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 4.515574074074074e-05,
+      "loss": 0.0033,
+      "step": 52340
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 4.515481481481482e-05,
+      "loss": 0.0052,
+      "step": 52350
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 4.515388888888889e-05,
+      "loss": 0.0045,
+      "step": 52360
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 4.515296296296296e-05,
+      "loss": 0.0042,
+      "step": 52370
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 4.5152037037037036e-05,
+      "loss": 0.0049,
+      "step": 52380
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 4.515111111111111e-05,
+      "loss": 0.0022,
+      "step": 52390
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 4.515018518518519e-05,
+      "loss": 0.0028,
+      "step": 52400
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 4.514925925925926e-05,
+      "loss": 0.0033,
+      "step": 52410
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 4.5148333333333336e-05,
+      "loss": 0.0032,
+      "step": 52420
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 4.514740740740741e-05,
+      "loss": 0.0037,
+      "step": 52430
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 4.5146481481481486e-05,
+      "loss": 0.0043,
+      "step": 52440
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 4.5145555555555555e-05,
+      "loss": 0.0053,
+      "step": 52450
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 4.514462962962963e-05,
+      "loss": 0.0029,
+      "step": 52460
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 4.5143703703703705e-05,
+      "loss": 0.0045,
+      "step": 52470
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 4.514277777777778e-05,
+      "loss": 0.0032,
+      "step": 52480
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 4.5141851851851855e-05,
+      "loss": 0.0038,
+      "step": 52490
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 4.514092592592592e-05,
+      "loss": 0.0036,
+      "step": 52500
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 4.5140000000000005e-05,
+      "loss": 0.0042,
+      "step": 52510
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 4.513907407407408e-05,
+      "loss": 0.0019,
+      "step": 52520
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 4.5138148148148155e-05,
+      "loss": 0.0038,
+      "step": 52530
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 4.513722222222222e-05,
+      "loss": 0.0032,
+      "step": 52540
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 4.51362962962963e-05,
+      "loss": 0.0047,
+      "step": 52550
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 4.513537037037037e-05,
+      "loss": 0.0071,
+      "step": 52560
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 4.513444444444445e-05,
+      "loss": 0.0033,
+      "step": 52570
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 4.513351851851852e-05,
+      "loss": 0.0062,
+      "step": 52580
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 4.513259259259259e-05,
+      "loss": 0.0071,
+      "step": 52590
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 4.513166666666667e-05,
+      "loss": 0.0032,
+      "step": 52600
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 4.513074074074075e-05,
+      "loss": 0.0028,
+      "step": 52610
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 4.512981481481482e-05,
+      "loss": 0.0038,
+      "step": 52620
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 4.512888888888889e-05,
+      "loss": 0.0061,
+      "step": 52630
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 4.512796296296297e-05,
+      "loss": 0.0031,
+      "step": 52640
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 4.512703703703704e-05,
+      "loss": 0.0027,
+      "step": 52650
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 4.512611111111111e-05,
+      "loss": 0.0053,
+      "step": 52660
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 4.5125185185185185e-05,
+      "loss": 0.0059,
+      "step": 52670
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 4.512425925925926e-05,
+      "loss": 0.005,
+      "step": 52680
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 4.5123333333333336e-05,
+      "loss": 0.0054,
+      "step": 52690
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 4.512240740740741e-05,
+      "loss": 0.005,
+      "step": 52700
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 4.512148148148148e-05,
+      "loss": 0.0037,
+      "step": 52710
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 4.512055555555556e-05,
+      "loss": 0.0037,
+      "step": 52720
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 4.5119629629629636e-05,
+      "loss": 0.0024,
+      "step": 52730
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 4.511870370370371e-05,
+      "loss": 0.0039,
+      "step": 52740
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 4.511777777777778e-05,
+      "loss": 0.0022,
+      "step": 52750
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 4.5116851851851854e-05,
+      "loss": 0.0058,
+      "step": 52760
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 4.511592592592593e-05,
+      "loss": 0.0037,
+      "step": 52770
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 4.5115000000000004e-05,
+      "loss": 0.0018,
+      "step": 52780
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 4.511407407407407e-05,
+      "loss": 0.0054,
+      "step": 52790
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 4.511314814814815e-05,
+      "loss": 0.005,
+      "step": 52800
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 4.511222222222222e-05,
+      "loss": 0.0042,
+      "step": 52810
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 4.51112962962963e-05,
+      "loss": 0.0038,
+      "step": 52820
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 4.511037037037037e-05,
+      "loss": 0.0031,
+      "step": 52830
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 4.510944444444445e-05,
+      "loss": 0.0049,
+      "step": 52840
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 4.510851851851852e-05,
+      "loss": 0.0032,
+      "step": 52850
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 4.51075925925926e-05,
+      "loss": 0.0083,
+      "step": 52860
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 4.5106666666666666e-05,
+      "loss": 0.0045,
+      "step": 52870
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 4.510574074074074e-05,
+      "loss": 0.0078,
+      "step": 52880
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 4.5104814814814816e-05,
+      "loss": 0.0067,
+      "step": 52890
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 4.510388888888889e-05,
+      "loss": 0.0047,
+      "step": 52900
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 4.5102962962962966e-05,
+      "loss": 0.0074,
+      "step": 52910
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 4.5102037037037035e-05,
+      "loss": 0.0052,
+      "step": 52920
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 4.510111111111111e-05,
+      "loss": 0.0059,
+      "step": 52930
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 4.510018518518519e-05,
+      "loss": 0.0067,
+      "step": 52940
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 4.509925925925927e-05,
+      "loss": 0.0059,
+      "step": 52950
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 4.5098333333333335e-05,
+      "loss": 0.0048,
+      "step": 52960
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 4.509740740740741e-05,
+      "loss": 0.0036,
+      "step": 52970
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 4.5096481481481485e-05,
+      "loss": 0.0032,
+      "step": 52980
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 4.509555555555556e-05,
+      "loss": 0.0036,
+      "step": 52990
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 4.509462962962963e-05,
+      "loss": 0.0048,
+      "step": 53000
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 4.5093703703703703e-05,
+      "loss": 0.007,
+      "step": 53010
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 4.509277777777778e-05,
+      "loss": 0.0041,
+      "step": 53020
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 4.5091851851851854e-05,
+      "loss": 0.0042,
+      "step": 53030
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 4.509092592592593e-05,
+      "loss": 0.008,
+      "step": 53040
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 4.5090000000000004e-05,
+      "loss": 0.0023,
+      "step": 53050
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 4.508907407407408e-05,
+      "loss": 0.0033,
+      "step": 53060
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 4.5088148148148154e-05,
+      "loss": 0.0031,
+      "step": 53070
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 4.508722222222222e-05,
+      "loss": 0.0037,
+      "step": 53080
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 4.50862962962963e-05,
+      "loss": 0.0041,
+      "step": 53090
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 4.508537037037037e-05,
+      "loss": 0.0039,
+      "step": 53100
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 4.508444444444445e-05,
+      "loss": 0.002,
+      "step": 53110
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 4.508351851851852e-05,
+      "loss": 0.003,
+      "step": 53120
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 4.508259259259259e-05,
+      "loss": 0.0031,
+      "step": 53130
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 4.5081666666666666e-05,
+      "loss": 0.0032,
+      "step": 53140
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 4.508074074074075e-05,
+      "loss": 0.0035,
+      "step": 53150
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 4.507981481481482e-05,
+      "loss": 0.0033,
+      "step": 53160
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.507888888888889e-05,
+      "loss": 0.004,
+      "step": 53170
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.5077962962962966e-05,
+      "loss": 0.0037,
+      "step": 53180
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.507703703703704e-05,
+      "loss": 0.0061,
+      "step": 53190
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.5076111111111116e-05,
+      "loss": 0.0026,
+      "step": 53200
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.5075185185185184e-05,
+      "loss": 0.0062,
+      "step": 53210
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.507425925925926e-05,
+      "loss": 0.0033,
+      "step": 53220
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.5073333333333334e-05,
+      "loss": 0.0052,
+      "step": 53230
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.507240740740741e-05,
+      "loss": 0.0048,
+      "step": 53240
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.5071481481481484e-05,
+      "loss": 0.0057,
+      "step": 53250
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.507055555555556e-05,
+      "loss": 0.0059,
+      "step": 53260
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.5069629629629635e-05,
+      "loss": 0.0021,
+      "step": 53270
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.506870370370371e-05,
+      "loss": 0.0045,
+      "step": 53280
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.506777777777778e-05,
+      "loss": 0.0051,
+      "step": 53290
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.506685185185185e-05,
+      "loss": 0.0047,
+      "step": 53300
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.506592592592593e-05,
+      "loss": 0.0035,
+      "step": 53310
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.5065e-05,
+      "loss": 0.0038,
+      "step": 53320
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.506407407407408e-05,
+      "loss": 0.0042,
+      "step": 53330
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.5063148148148146e-05,
+      "loss": 0.0035,
+      "step": 53340
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.506222222222222e-05,
+      "loss": 0.0034,
+      "step": 53350
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.5061296296296297e-05,
+      "loss": 0.003,
+      "step": 53360
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.506037037037038e-05,
+      "loss": 0.003,
+      "step": 53370
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 4.505944444444445e-05,
+      "loss": 0.0045,
+      "step": 53380
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 4.505851851851852e-05,
+      "loss": 0.0058,
+      "step": 53390
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 4.50575925925926e-05,
+      "loss": 0.004,
+      "step": 53400
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 4.505666666666667e-05,
+      "loss": 0.005,
+      "step": 53410
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 4.505574074074074e-05,
+      "loss": 0.0045,
+      "step": 53420
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 4.5054814814814815e-05,
+      "loss": 0.0049,
+      "step": 53430
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 4.505388888888889e-05,
+      "loss": 0.0022,
+      "step": 53440
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 4.5052962962962965e-05,
+      "loss": 0.005,
+      "step": 53450
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 4.505203703703704e-05,
+      "loss": 0.0054,
+      "step": 53460
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 4.505111111111111e-05,
+      "loss": 0.005,
+      "step": 53470
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 4.505018518518519e-05,
+      "loss": 0.0037,
+      "step": 53480
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 4.5049259259259265e-05,
+      "loss": 0.0054,
+      "step": 53490
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 4.5048333333333334e-05,
+      "loss": 0.0025,
+      "step": 53500
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 4.504740740740741e-05,
+      "loss": 0.005,
+      "step": 53510
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 4.5046481481481484e-05,
+      "loss": 0.0022,
+      "step": 53520
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 4.504555555555556e-05,
+      "loss": 0.0026,
+      "step": 53530
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 4.5044629629629634e-05,
+      "loss": 0.008,
+      "step": 53540
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 4.50437037037037e-05,
+      "loss": 0.0056,
+      "step": 53550
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 4.504277777777778e-05,
+      "loss": 0.0045,
+      "step": 53560
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 4.504185185185185e-05,
+      "loss": 0.0041,
+      "step": 53570
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 4.5040925925925934e-05,
+      "loss": 0.0047,
+      "step": 53580
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 4.504e-05,
+      "loss": 0.0035,
+      "step": 53590
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 4.503907407407408e-05,
+      "loss": 0.0056,
+      "step": 53600
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 4.503814814814815e-05,
+      "loss": 0.0029,
+      "step": 53610
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 4.503722222222223e-05,
+      "loss": 0.0047,
+      "step": 53620
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 4.5036296296296296e-05,
+      "loss": 0.0021,
+      "step": 53630
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 4.503537037037037e-05,
+      "loss": 0.0056,
+      "step": 53640
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 4.5034444444444446e-05,
+      "loss": 0.0041,
+      "step": 53650
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 4.503351851851852e-05,
+      "loss": 0.0044,
+      "step": 53660
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 4.5032592592592596e-05,
+      "loss": 0.0038,
+      "step": 53670
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 4.5031666666666664e-05,
+      "loss": 0.0047,
+      "step": 53680
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 4.5030740740740746e-05,
+      "loss": 0.003,
+      "step": 53690
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 4.502981481481482e-05,
+      "loss": 0.0025,
+      "step": 53700
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 4.502888888888889e-05,
+      "loss": 0.0036,
+      "step": 53710
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 4.5027962962962965e-05,
+      "loss": 0.0045,
+      "step": 53720
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 4.502703703703704e-05,
+      "loss": 0.004,
+      "step": 53730
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 4.5026111111111115e-05,
+      "loss": 0.0063,
+      "step": 53740
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 4.502518518518519e-05,
+      "loss": 0.0023,
+      "step": 53750
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 4.502425925925926e-05,
+      "loss": 0.0056,
+      "step": 53760
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 4.502333333333333e-05,
+      "loss": 0.0052,
+      "step": 53770
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 4.502240740740741e-05,
+      "loss": 0.0024,
+      "step": 53780
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 4.502148148148148e-05,
+      "loss": 0.0031,
+      "step": 53790
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 4.502055555555556e-05,
+      "loss": 0.0045,
+      "step": 53800
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 4.501962962962963e-05,
+      "loss": 0.0049,
+      "step": 53810
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 4.501870370370371e-05,
+      "loss": 0.005,
+      "step": 53820
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 4.5017777777777783e-05,
+      "loss": 0.0036,
+      "step": 53830
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 4.501685185185185e-05,
+      "loss": 0.0039,
+      "step": 53840
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 4.501592592592593e-05,
+      "loss": 0.0047,
+      "step": 53850
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 4.5015e-05,
+      "loss": 0.0032,
+      "step": 53860
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 4.501407407407408e-05,
+      "loss": 0.0054,
+      "step": 53870
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 4.501314814814815e-05,
+      "loss": 0.0038,
+      "step": 53880
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 4.501222222222222e-05,
+      "loss": 0.0069,
+      "step": 53890
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 4.5011296296296295e-05,
+      "loss": 0.0034,
+      "step": 53900
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 4.501037037037038e-05,
+      "loss": 0.0029,
+      "step": 53910
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 4.5009444444444445e-05,
+      "loss": 0.006,
+      "step": 53920
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 4.500851851851852e-05,
+      "loss": 0.0052,
+      "step": 53930
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 4.5007592592592596e-05,
+      "loss": 0.0051,
+      "step": 53940
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 4.500666666666667e-05,
+      "loss": 0.0036,
+      "step": 53950
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 4.5005740740740746e-05,
+      "loss": 0.0044,
+      "step": 53960
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 4.5004814814814814e-05,
+      "loss": 0.0034,
+      "step": 53970
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 4.500388888888889e-05,
+      "loss": 0.0035,
+      "step": 53980
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 4.5002962962962964e-05,
+      "loss": 0.0026,
+      "step": 53990
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 4.500203703703704e-05,
+      "loss": 0.0049,
+      "step": 54000
+    },
+    {
+      "epoch": 10.0,
+      "eval_loss": 0.013164997100830078,
+      "eval_runtime": 153.5507,
+      "eval_samples_per_second": 25.985,
+      "eval_steps_per_second": 3.25,
+      "step": 54000
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 4.500111111111111e-05,
+      "loss": 0.0021,
+      "step": 54010
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 4.500018518518519e-05,
+      "loss": 0.0044,
+      "step": 54020
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 4.4999259259259264e-05,
+      "loss": 0.0039,
+      "step": 54030
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 4.499833333333334e-05,
+      "loss": 0.0032,
+      "step": 54040
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 4.499740740740741e-05,
+      "loss": 0.0023,
+      "step": 54050
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 4.499648148148148e-05,
+      "loss": 0.0044,
+      "step": 54060
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 4.499555555555556e-05,
+      "loss": 0.0031,
+      "step": 54070
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 4.499462962962963e-05,
+      "loss": 0.0029,
+      "step": 54080
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.499370370370371e-05,
+      "loss": 0.0028,
+      "step": 54090
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.4992777777777776e-05,
+      "loss": 0.0028,
+      "step": 54100
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.499185185185185e-05,
+      "loss": 0.0034,
+      "step": 54110
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.499092592592593e-05,
+      "loss": 0.0029,
+      "step": 54120
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.499e-05,
+      "loss": 0.0026,
+      "step": 54130
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.4989074074074076e-05,
+      "loss": 0.0034,
+      "step": 54140
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.498814814814815e-05,
+      "loss": 0.0018,
+      "step": 54150
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.4987222222222226e-05,
+      "loss": 0.0034,
+      "step": 54160
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.49862962962963e-05,
+      "loss": 0.002,
+      "step": 54170
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.498537037037037e-05,
+      "loss": 0.0027,
+      "step": 54180
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.4984444444444445e-05,
+      "loss": 0.0035,
+      "step": 54190
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.498351851851852e-05,
+      "loss": 0.0037,
+      "step": 54200
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.4982592592592595e-05,
+      "loss": 0.0031,
+      "step": 54210
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.498166666666666e-05,
+      "loss": 0.0037,
+      "step": 54220
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.4980740740740745e-05,
+      "loss": 0.0032,
+      "step": 54230
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.497981481481482e-05,
+      "loss": 0.0023,
+      "step": 54240
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.4978888888888895e-05,
+      "loss": 0.0049,
+      "step": 54250
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.4977962962962963e-05,
+      "loss": 0.0026,
+      "step": 54260
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.497703703703704e-05,
+      "loss": 0.0053,
+      "step": 54270
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.4976111111111114e-05,
+      "loss": 0.0025,
+      "step": 54280
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.497518518518519e-05,
+      "loss": 0.0019,
+      "step": 54290
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.4974259259259264e-05,
+      "loss": 0.0025,
+      "step": 54300
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.497333333333333e-05,
+      "loss": 0.003,
+      "step": 54310
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.497240740740741e-05,
+      "loss": 0.0029,
+      "step": 54320
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.497148148148148e-05,
+      "loss": 0.0037,
+      "step": 54330
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.497055555555556e-05,
+      "loss": 0.0041,
+      "step": 54340
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.496962962962963e-05,
+      "loss": 0.0028,
+      "step": 54350
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.496870370370371e-05,
+      "loss": 0.0041,
+      "step": 54360
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.496777777777778e-05,
+      "loss": 0.0047,
+      "step": 54370
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.496685185185186e-05,
+      "loss": 0.0024,
+      "step": 54380
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.4965925925925926e-05,
+      "loss": 0.0025,
+      "step": 54390
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.4965e-05,
+      "loss": 0.0035,
+      "step": 54400
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.4964074074074076e-05,
+      "loss": 0.0017,
+      "step": 54410
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.496314814814815e-05,
+      "loss": 0.0036,
+      "step": 54420
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.496222222222222e-05,
+      "loss": 0.0026,
+      "step": 54430
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.4961296296296294e-05,
+      "loss": 0.0056,
+      "step": 54440
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.4960462962962966e-05,
+      "loss": 0.0076,
+      "step": 54450
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.495953703703704e-05,
+      "loss": 0.0026,
+      "step": 54460
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.4958611111111116e-05,
+      "loss": 0.0021,
+      "step": 54470
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.495768518518519e-05,
+      "loss": 0.0042,
+      "step": 54480
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.495675925925926e-05,
+      "loss": 0.0055,
+      "step": 54490
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.4955833333333334e-05,
+      "loss": 0.0062,
+      "step": 54500
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.495490740740741e-05,
+      "loss": 0.0041,
+      "step": 54510
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.4953981481481484e-05,
+      "loss": 0.0078,
+      "step": 54520
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.495305555555556e-05,
+      "loss": 0.004,
+      "step": 54530
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.495212962962963e-05,
+      "loss": 0.0053,
+      "step": 54540
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.49512037037037e-05,
+      "loss": 0.0053,
+      "step": 54550
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.495027777777778e-05,
+      "loss": 0.0047,
+      "step": 54560
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.494935185185186e-05,
+      "loss": 0.0041,
+      "step": 54570
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.494842592592593e-05,
+      "loss": 0.003,
+      "step": 54580
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.49475e-05,
+      "loss": 0.0038,
+      "step": 54590
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.494657407407408e-05,
+      "loss": 0.0036,
+      "step": 54600
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.494564814814815e-05,
+      "loss": 0.0027,
+      "step": 54610
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.494472222222222e-05,
+      "loss": 0.0029,
+      "step": 54620
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.4943796296296296e-05,
+      "loss": 0.0041,
+      "step": 54630
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.494287037037037e-05,
+      "loss": 0.0036,
+      "step": 54640
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.4941944444444446e-05,
+      "loss": 0.0024,
+      "step": 54650
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.494101851851852e-05,
+      "loss": 0.0028,
+      "step": 54660
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.4940092592592597e-05,
+      "loss": 0.0026,
+      "step": 54670
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.493916666666667e-05,
+      "loss": 0.0078,
+      "step": 54680
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.493824074074075e-05,
+      "loss": 0.004,
+      "step": 54690
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.4937314814814815e-05,
+      "loss": 0.0015,
+      "step": 54700
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.493638888888889e-05,
+      "loss": 0.0026,
+      "step": 54710
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.4935462962962965e-05,
+      "loss": 0.0039,
+      "step": 54720
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.493453703703704e-05,
+      "loss": 0.0047,
+      "step": 54730
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.4933611111111115e-05,
+      "loss": 0.0045,
+      "step": 54740
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.4932685185185183e-05,
+      "loss": 0.0029,
+      "step": 54750
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.493175925925926e-05,
+      "loss": 0.0039,
+      "step": 54760
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.4930833333333334e-05,
+      "loss": 0.0037,
+      "step": 54770
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.4929907407407415e-05,
+      "loss": 0.004,
+      "step": 54780
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.4928981481481484e-05,
+      "loss": 0.0032,
+      "step": 54790
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.492805555555556e-05,
+      "loss": 0.0064,
+      "step": 54800
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.4927129629629634e-05,
+      "loss": 0.0036,
+      "step": 54810
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.492620370370371e-05,
+      "loss": 0.0039,
+      "step": 54820
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.492527777777778e-05,
+      "loss": 0.0108,
+      "step": 54830
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.492435185185185e-05,
+      "loss": 0.0042,
+      "step": 54840
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.492342592592593e-05,
+      "loss": 0.0032,
+      "step": 54850
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.49225e-05,
+      "loss": 0.0043,
+      "step": 54860
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.492157407407408e-05,
+      "loss": 0.004,
+      "step": 54870
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.4920648148148146e-05,
+      "loss": 0.0024,
+      "step": 54880
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.491972222222223e-05,
+      "loss": 0.002,
+      "step": 54890
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.49187962962963e-05,
+      "loss": 0.0049,
+      "step": 54900
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.491787037037037e-05,
+      "loss": 0.0031,
+      "step": 54910
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.4916944444444446e-05,
+      "loss": 0.0024,
+      "step": 54920
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.491601851851852e-05,
+      "loss": 0.0028,
+      "step": 54930
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.4915092592592596e-05,
+      "loss": 0.0018,
+      "step": 54940
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.491416666666667e-05,
+      "loss": 0.0053,
+      "step": 54950
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.491324074074074e-05,
+      "loss": 0.0057,
+      "step": 54960
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.4912314814814814e-05,
+      "loss": 0.0036,
+      "step": 54970
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.491138888888889e-05,
+      "loss": 0.0053,
+      "step": 54980
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.4910462962962964e-05,
+      "loss": 0.0078,
+      "step": 54990
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.490953703703704e-05,
+      "loss": 0.002,
+      "step": 55000
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.4908611111111115e-05,
+      "loss": 0.0032,
+      "step": 55010
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.490768518518519e-05,
+      "loss": 0.0028,
+      "step": 55020
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.4906759259259265e-05,
+      "loss": 0.006,
+      "step": 55030
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.490583333333333e-05,
+      "loss": 0.0029,
+      "step": 55040
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.490490740740741e-05,
+      "loss": 0.003,
+      "step": 55050
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.490398148148148e-05,
+      "loss": 0.0037,
+      "step": 55060
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.490305555555556e-05,
+      "loss": 0.0032,
+      "step": 55070
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.490212962962963e-05,
+      "loss": 0.0027,
+      "step": 55080
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.49012037037037e-05,
+      "loss": 0.0031,
+      "step": 55090
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.4900277777777777e-05,
+      "loss": 0.005,
+      "step": 55100
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.489935185185186e-05,
+      "loss": 0.003,
+      "step": 55110
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.489842592592593e-05,
+      "loss": 0.0025,
+      "step": 55120
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.48975e-05,
+      "loss": 0.0028,
+      "step": 55130
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.489657407407408e-05,
+      "loss": 0.0033,
+      "step": 55140
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.489564814814815e-05,
+      "loss": 0.0014,
+      "step": 55150
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.489472222222223e-05,
+      "loss": 0.0032,
+      "step": 55160
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.4893796296296295e-05,
+      "loss": 0.0038,
+      "step": 55170
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.489287037037037e-05,
+      "loss": 0.0019,
+      "step": 55180
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.4891944444444445e-05,
+      "loss": 0.0064,
+      "step": 55190
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.489101851851852e-05,
+      "loss": 0.0024,
+      "step": 55200
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.4890092592592595e-05,
+      "loss": 0.006,
+      "step": 55210
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.488916666666667e-05,
+      "loss": 0.0016,
+      "step": 55220
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.4888240740740745e-05,
+      "loss": 0.0048,
+      "step": 55230
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.488731481481482e-05,
+      "loss": 0.0045,
+      "step": 55240
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.488638888888889e-05,
+      "loss": 0.0029,
+      "step": 55250
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.4885462962962964e-05,
+      "loss": 0.0042,
+      "step": 55260
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.488453703703704e-05,
+      "loss": 0.0044,
+      "step": 55270
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.4883611111111114e-05,
+      "loss": 0.0041,
+      "step": 55280
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.488268518518519e-05,
+      "loss": 0.0023,
+      "step": 55290
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.488175925925926e-05,
+      "loss": 0.0042,
+      "step": 55300
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.488083333333333e-05,
+      "loss": 0.003,
+      "step": 55310
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.4879907407407414e-05,
+      "loss": 0.0031,
+      "step": 55320
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.487898148148148e-05,
+      "loss": 0.005,
+      "step": 55330
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.487805555555556e-05,
+      "loss": 0.0042,
+      "step": 55340
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.487712962962963e-05,
+      "loss": 0.004,
+      "step": 55350
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.487620370370371e-05,
+      "loss": 0.0023,
+      "step": 55360
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.487527777777778e-05,
+      "loss": 0.0032,
+      "step": 55370
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.487435185185185e-05,
+      "loss": 0.0039,
+      "step": 55380
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.4873425925925926e-05,
+      "loss": 0.0035,
+      "step": 55390
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.48725e-05,
+      "loss": 0.003,
+      "step": 55400
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.4871574074074076e-05,
+      "loss": 0.0037,
+      "step": 55410
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.4870648148148144e-05,
+      "loss": 0.0035,
+      "step": 55420
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.4869722222222226e-05,
+      "loss": 0.0057,
+      "step": 55430
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.48687962962963e-05,
+      "loss": 0.002,
+      "step": 55440
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.4867870370370376e-05,
+      "loss": 0.0081,
+      "step": 55450
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.4866944444444445e-05,
+      "loss": 0.0034,
+      "step": 55460
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.486601851851852e-05,
+      "loss": 0.0066,
+      "step": 55470
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.4865092592592595e-05,
+      "loss": 0.0046,
+      "step": 55480
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.486416666666667e-05,
+      "loss": 0.005,
+      "step": 55490
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.4863240740740745e-05,
+      "loss": 0.0032,
+      "step": 55500
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.486231481481481e-05,
+      "loss": 0.0051,
+      "step": 55510
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.486138888888889e-05,
+      "loss": 0.0021,
+      "step": 55520
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.486046296296296e-05,
+      "loss": 0.0065,
+      "step": 55530
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.485953703703704e-05,
+      "loss": 0.0033,
+      "step": 55540
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.485861111111111e-05,
+      "loss": 0.0025,
+      "step": 55550
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.485768518518519e-05,
+      "loss": 0.0035,
+      "step": 55560
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.4856759259259263e-05,
+      "loss": 0.0025,
+      "step": 55570
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.485583333333334e-05,
+      "loss": 0.004,
+      "step": 55580
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.485490740740741e-05,
+      "loss": 0.0037,
+      "step": 55590
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.485398148148148e-05,
+      "loss": 0.0044,
+      "step": 55600
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.485305555555556e-05,
+      "loss": 0.0036,
+      "step": 55610
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.485212962962963e-05,
+      "loss": 0.0018,
+      "step": 55620
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.48512037037037e-05,
+      "loss": 0.0026,
+      "step": 55630
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.4850277777777775e-05,
+      "loss": 0.0022,
+      "step": 55640
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.484935185185186e-05,
+      "loss": 0.0075,
+      "step": 55650
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.484842592592593e-05,
+      "loss": 0.0052,
+      "step": 55660
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.48475e-05,
+      "loss": 0.0027,
+      "step": 55670
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.4846574074074076e-05,
+      "loss": 0.0072,
+      "step": 55680
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.484564814814815e-05,
+      "loss": 0.0047,
+      "step": 55690
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.4844722222222226e-05,
+      "loss": 0.0032,
+      "step": 55700
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.48437962962963e-05,
+      "loss": 0.0048,
+      "step": 55710
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.484287037037037e-05,
+      "loss": 0.0089,
+      "step": 55720
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.4841944444444444e-05,
+      "loss": 0.0029,
+      "step": 55730
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.484101851851852e-05,
+      "loss": 0.0041,
+      "step": 55740
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.4840092592592594e-05,
+      "loss": 0.0026,
+      "step": 55750
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.483916666666667e-05,
+      "loss": 0.004,
+      "step": 55760
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.4838240740740744e-05,
+      "loss": 0.0054,
+      "step": 55770
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.483731481481482e-05,
+      "loss": 0.0023,
+      "step": 55780
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.4836388888888894e-05,
+      "loss": 0.0028,
+      "step": 55790
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.483546296296296e-05,
+      "loss": 0.0039,
+      "step": 55800
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.483453703703704e-05,
+      "loss": 0.0021,
+      "step": 55810
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.483361111111111e-05,
+      "loss": 0.0024,
+      "step": 55820
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.483268518518519e-05,
+      "loss": 0.0052,
+      "step": 55830
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.4831759259259256e-05,
+      "loss": 0.0037,
+      "step": 55840
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.483083333333333e-05,
+      "loss": 0.0025,
+      "step": 55850
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.482990740740741e-05,
+      "loss": 0.003,
+      "step": 55860
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.482898148148149e-05,
+      "loss": 0.0038,
+      "step": 55870
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.4828055555555556e-05,
+      "loss": 0.003,
+      "step": 55880
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.482712962962963e-05,
+      "loss": 0.0051,
+      "step": 55890
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.4826203703703706e-05,
+      "loss": 0.0037,
+      "step": 55900
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.482527777777778e-05,
+      "loss": 0.0038,
+      "step": 55910
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.4824351851851857e-05,
+      "loss": 0.0036,
+      "step": 55920
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.4823425925925925e-05,
+      "loss": 0.0028,
+      "step": 55930
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.48225e-05,
+      "loss": 0.0032,
+      "step": 55940
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.4821574074074075e-05,
+      "loss": 0.0034,
+      "step": 55950
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.482064814814815e-05,
+      "loss": 0.0032,
+      "step": 55960
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.4819722222222225e-05,
+      "loss": 0.0039,
+      "step": 55970
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.48187962962963e-05,
+      "loss": 0.0052,
+      "step": 55980
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.4817870370370375e-05,
+      "loss": 0.0038,
+      "step": 55990
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.481694444444445e-05,
+      "loss": 0.0036,
+      "step": 56000
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.481601851851852e-05,
+      "loss": 0.0031,
+      "step": 56010
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.4815092592592594e-05,
+      "loss": 0.0029,
+      "step": 56020
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.481416666666667e-05,
+      "loss": 0.0052,
+      "step": 56030
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.4813240740740744e-05,
+      "loss": 0.0025,
+      "step": 56040
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.481231481481481e-05,
+      "loss": 0.0047,
+      "step": 56050
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.481138888888889e-05,
+      "loss": 0.0046,
+      "step": 56060
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.481046296296296e-05,
+      "loss": 0.0036,
+      "step": 56070
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.4809537037037044e-05,
+      "loss": 0.0038,
+      "step": 56080
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.480861111111111e-05,
+      "loss": 0.0039,
+      "step": 56090
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.480768518518519e-05,
+      "loss": 0.0057,
+      "step": 56100
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.480675925925926e-05,
+      "loss": 0.0051,
+      "step": 56110
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.480583333333334e-05,
+      "loss": 0.0029,
+      "step": 56120
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.480490740740741e-05,
+      "loss": 0.0022,
+      "step": 56130
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.480398148148148e-05,
+      "loss": 0.0041,
+      "step": 56140
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.4803055555555556e-05,
+      "loss": 0.0046,
+      "step": 56150
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.480212962962963e-05,
+      "loss": 0.0034,
+      "step": 56160
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.4801203703703706e-05,
+      "loss": 0.0043,
+      "step": 56170
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.480027777777778e-05,
+      "loss": 0.0062,
+      "step": 56180
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.4799351851851856e-05,
+      "loss": 0.0044,
+      "step": 56190
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.479842592592593e-05,
+      "loss": 0.0035,
+      "step": 56200
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.4797500000000006e-05,
+      "loss": 0.0029,
+      "step": 56210
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.4796574074074074e-05,
+      "loss": 0.0036,
+      "step": 56220
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.479564814814815e-05,
+      "loss": 0.0028,
+      "step": 56230
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.4794722222222224e-05,
+      "loss": 0.0039,
+      "step": 56240
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.47937962962963e-05,
+      "loss": 0.0031,
+      "step": 56250
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.479287037037037e-05,
+      "loss": 0.0028,
+      "step": 56260
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.479194444444444e-05,
+      "loss": 0.0034,
+      "step": 56270
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.479101851851852e-05,
+      "loss": 0.0045,
+      "step": 56280
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.47900925925926e-05,
+      "loss": 0.0028,
+      "step": 56290
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.478916666666667e-05,
+      "loss": 0.004,
+      "step": 56300
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.478824074074074e-05,
+      "loss": 0.0028,
+      "step": 56310
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.478731481481482e-05,
+      "loss": 0.0038,
+      "step": 56320
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.478638888888889e-05,
+      "loss": 0.0041,
+      "step": 56330
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.478546296296297e-05,
+      "loss": 0.0054,
+      "step": 56340
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.4784537037037036e-05,
+      "loss": 0.0041,
+      "step": 56350
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.478361111111111e-05,
+      "loss": 0.0045,
+      "step": 56360
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.4782685185185187e-05,
+      "loss": 0.0057,
+      "step": 56370
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.478175925925926e-05,
+      "loss": 0.0052,
+      "step": 56380
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.478083333333333e-05,
+      "loss": 0.0052,
+      "step": 56390
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.477990740740741e-05,
+      "loss": 0.0038,
+      "step": 56400
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.477898148148149e-05,
+      "loss": 0.0049,
+      "step": 56410
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.477805555555556e-05,
+      "loss": 0.0045,
+      "step": 56420
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.477712962962963e-05,
+      "loss": 0.0036,
+      "step": 56430
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.4776203703703705e-05,
+      "loss": 0.0033,
+      "step": 56440
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.477527777777778e-05,
+      "loss": 0.0017,
+      "step": 56450
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.4774351851851855e-05,
+      "loss": 0.0038,
+      "step": 56460
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.477342592592593e-05,
+      "loss": 0.0032,
+      "step": 56470
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.47725e-05,
+      "loss": 0.0026,
+      "step": 56480
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.4771574074074074e-05,
+      "loss": 0.0027,
+      "step": 56490
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.477064814814815e-05,
+      "loss": 0.005,
+      "step": 56500
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.4769722222222224e-05,
+      "loss": 0.0033,
+      "step": 56510
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.47687962962963e-05,
+      "loss": 0.003,
+      "step": 56520
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.4767870370370374e-05,
+      "loss": 0.0037,
+      "step": 56530
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.476694444444445e-05,
+      "loss": 0.0025,
+      "step": 56540
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.4766018518518524e-05,
+      "loss": 0.0043,
+      "step": 56550
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.476509259259259e-05,
+      "loss": 0.0044,
+      "step": 56560
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.476416666666667e-05,
+      "loss": 0.0032,
+      "step": 56570
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.476324074074074e-05,
+      "loss": 0.0041,
+      "step": 56580
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.476231481481482e-05,
+      "loss": 0.003,
+      "step": 56590
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.4761388888888886e-05,
+      "loss": 0.0034,
+      "step": 56600
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.476046296296296e-05,
+      "loss": 0.0028,
+      "step": 56610
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.475953703703704e-05,
+      "loss": 0.0065,
+      "step": 56620
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.475861111111112e-05,
+      "loss": 0.0023,
+      "step": 56630
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.4757685185185186e-05,
+      "loss": 0.0042,
+      "step": 56640
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.475685185185186e-05,
+      "loss": 0.0053,
+      "step": 56650
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.4755925925925926e-05,
+      "loss": 0.0054,
+      "step": 56660
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.4755e-05,
+      "loss": 0.004,
+      "step": 56670
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.4754074074074076e-05,
+      "loss": 0.0031,
+      "step": 56680
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.475314814814815e-05,
+      "loss": 0.0037,
+      "step": 56690
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.4752222222222226e-05,
+      "loss": 0.0055,
+      "step": 56700
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.4751296296296294e-05,
+      "loss": 0.0024,
+      "step": 56710
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.475037037037037e-05,
+      "loss": 0.0047,
+      "step": 56720
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.4749444444444445e-05,
+      "loss": 0.0024,
+      "step": 56730
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.4748518518518526e-05,
+      "loss": 0.0033,
+      "step": 56740
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.4747592592592595e-05,
+      "loss": 0.0048,
+      "step": 56750
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.474666666666667e-05,
+      "loss": 0.0026,
+      "step": 56760
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.4745740740740745e-05,
+      "loss": 0.0039,
+      "step": 56770
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.474481481481482e-05,
+      "loss": 0.0028,
+      "step": 56780
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.474388888888889e-05,
+      "loss": 0.0055,
+      "step": 56790
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.474296296296296e-05,
+      "loss": 0.0031,
+      "step": 56800
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.474203703703704e-05,
+      "loss": 0.0043,
+      "step": 56810
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.474111111111111e-05,
+      "loss": 0.0017,
+      "step": 56820
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.474018518518519e-05,
+      "loss": 0.0032,
+      "step": 56830
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.473925925925926e-05,
+      "loss": 0.0046,
+      "step": 56840
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.473833333333334e-05,
+      "loss": 0.0054,
+      "step": 56850
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.4737407407407413e-05,
+      "loss": 0.0046,
+      "step": 56860
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.473648148148148e-05,
+      "loss": 0.0049,
+      "step": 56870
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.473555555555556e-05,
+      "loss": 0.0072,
+      "step": 56880
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.473462962962963e-05,
+      "loss": 0.0052,
+      "step": 56890
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.473370370370371e-05,
+      "loss": 0.0022,
+      "step": 56900
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.473277777777778e-05,
+      "loss": 0.0074,
+      "step": 56910
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.473185185185185e-05,
+      "loss": 0.0025,
+      "step": 56920
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.4730925925925925e-05,
+      "loss": 0.0038,
+      "step": 56930
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.473e-05,
+      "loss": 0.0029,
+      "step": 56940
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.472907407407408e-05,
+      "loss": 0.0043,
+      "step": 56950
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.472814814814815e-05,
+      "loss": 0.005,
+      "step": 56960
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.4727222222222226e-05,
+      "loss": 0.003,
+      "step": 56970
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.47262962962963e-05,
+      "loss": 0.004,
+      "step": 56980
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.4725370370370376e-05,
+      "loss": 0.0027,
+      "step": 56990
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.4724444444444444e-05,
+      "loss": 0.0038,
+      "step": 57000
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.472351851851852e-05,
+      "loss": 0.0031,
+      "step": 57010
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.4722592592592594e-05,
+      "loss": 0.0035,
+      "step": 57020
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.472166666666667e-05,
+      "loss": 0.0027,
+      "step": 57030
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.4720740740740744e-05,
+      "loss": 0.0039,
+      "step": 57040
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.471981481481481e-05,
+      "loss": 0.0034,
+      "step": 57050
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.4718888888888894e-05,
+      "loss": 0.0028,
+      "step": 57060
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.471796296296297e-05,
+      "loss": 0.0025,
+      "step": 57070
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.471703703703704e-05,
+      "loss": 0.0027,
+      "step": 57080
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.471611111111111e-05,
+      "loss": 0.0052,
+      "step": 57090
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.471518518518519e-05,
+      "loss": 0.0052,
+      "step": 57100
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.471425925925926e-05,
+      "loss": 0.0022,
+      "step": 57110
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.471333333333334e-05,
+      "loss": 0.004,
+      "step": 57120
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.4712407407407406e-05,
+      "loss": 0.0027,
+      "step": 57130
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.471148148148148e-05,
+      "loss": 0.0052,
+      "step": 57140
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.4710555555555556e-05,
+      "loss": 0.0025,
+      "step": 57150
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.470962962962963e-05,
+      "loss": 0.005,
+      "step": 57160
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.4708703703703706e-05,
+      "loss": 0.0046,
+      "step": 57170
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.470777777777778e-05,
+      "loss": 0.0032,
+      "step": 57180
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.4706851851851856e-05,
+      "loss": 0.0067,
+      "step": 57190
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.470592592592593e-05,
+      "loss": 0.0038,
+      "step": 57200
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.4705e-05,
+      "loss": 0.0026,
+      "step": 57210
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.4704074074074075e-05,
+      "loss": 0.0042,
+      "step": 57220
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.470314814814815e-05,
+      "loss": 0.0026,
+      "step": 57230
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.4702222222222225e-05,
+      "loss": 0.0041,
+      "step": 57240
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.47012962962963e-05,
+      "loss": 0.0044,
+      "step": 57250
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.470037037037037e-05,
+      "loss": 0.0022,
+      "step": 57260
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.469944444444445e-05,
+      "loss": 0.0023,
+      "step": 57270
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.4698518518518525e-05,
+      "loss": 0.0042,
+      "step": 57280
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.4697592592592593e-05,
+      "loss": 0.0031,
+      "step": 57290
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.469666666666667e-05,
+      "loss": 0.0049,
+      "step": 57300
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.4695740740740744e-05,
+      "loss": 0.0014,
+      "step": 57310
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.469481481481482e-05,
+      "loss": 0.0029,
+      "step": 57320
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.4693888888888894e-05,
+      "loss": 0.0031,
+      "step": 57330
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.469296296296296e-05,
+      "loss": 0.0028,
+      "step": 57340
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.469203703703704e-05,
+      "loss": 0.0024,
+      "step": 57350
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.469111111111111e-05,
+      "loss": 0.0024,
+      "step": 57360
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.469018518518519e-05,
+      "loss": 0.005,
+      "step": 57370
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.468925925925926e-05,
+      "loss": 0.0035,
+      "step": 57380
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.468833333333334e-05,
+      "loss": 0.0028,
+      "step": 57390
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.468740740740741e-05,
+      "loss": 0.0051,
+      "step": 57400
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.468648148148149e-05,
+      "loss": 0.0031,
+      "step": 57410
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.4685555555555556e-05,
+      "loss": 0.0008,
+      "step": 57420
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.468462962962963e-05,
+      "loss": 0.0045,
+      "step": 57430
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.4683703703703706e-05,
+      "loss": 0.0037,
+      "step": 57440
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.468277777777778e-05,
+      "loss": 0.0021,
+      "step": 57450
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.4681851851851856e-05,
+      "loss": 0.0031,
+      "step": 57460
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.4680925925925924e-05,
+      "loss": 0.0023,
+      "step": 57470
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.468e-05,
+      "loss": 0.0051,
+      "step": 57480
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.467907407407408e-05,
+      "loss": 0.0045,
+      "step": 57490
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.467814814814815e-05,
+      "loss": 0.0037,
+      "step": 57500
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.4677222222222224e-05,
+      "loss": 0.0045,
+      "step": 57510
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.46762962962963e-05,
+      "loss": 0.004,
+      "step": 57520
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.4675370370370374e-05,
+      "loss": 0.0025,
+      "step": 57530
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.467444444444445e-05,
+      "loss": 0.0035,
+      "step": 57540
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.467351851851852e-05,
+      "loss": 0.0038,
+      "step": 57550
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.467259259259259e-05,
+      "loss": 0.0039,
+      "step": 57560
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.467166666666667e-05,
+      "loss": 0.0034,
+      "step": 57570
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.467074074074074e-05,
+      "loss": 0.0041,
+      "step": 57580
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.466981481481481e-05,
+      "loss": 0.0041,
+      "step": 57590
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.466888888888889e-05,
+      "loss": 0.006,
+      "step": 57600
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.466796296296297e-05,
+      "loss": 0.0037,
+      "step": 57610
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.466703703703704e-05,
+      "loss": 0.0033,
+      "step": 57620
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.466611111111111e-05,
+      "loss": 0.0045,
+      "step": 57630
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.4665185185185186e-05,
+      "loss": 0.0032,
+      "step": 57640
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.466425925925926e-05,
+      "loss": 0.0043,
+      "step": 57650
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.4663333333333337e-05,
+      "loss": 0.0065,
+      "step": 57660
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.466240740740741e-05,
+      "loss": 0.0039,
+      "step": 57670
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.466148148148148e-05,
+      "loss": 0.003,
+      "step": 57680
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.4660555555555555e-05,
+      "loss": 0.0031,
+      "step": 57690
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.465962962962963e-05,
+      "loss": 0.0054,
+      "step": 57700
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.4658703703703705e-05,
+      "loss": 0.0046,
+      "step": 57710
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.465777777777778e-05,
+      "loss": 0.0036,
+      "step": 57720
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.4656851851851855e-05,
+      "loss": 0.0019,
+      "step": 57730
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.465592592592593e-05,
+      "loss": 0.0034,
+      "step": 57740
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.4655000000000005e-05,
+      "loss": 0.0036,
+      "step": 57750
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.4654074074074074e-05,
+      "loss": 0.004,
+      "step": 57760
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.465314814814815e-05,
+      "loss": 0.0035,
+      "step": 57770
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.4652222222222224e-05,
+      "loss": 0.0028,
+      "step": 57780
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.46512962962963e-05,
+      "loss": 0.0069,
+      "step": 57790
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.465037037037037e-05,
+      "loss": 0.0024,
+      "step": 57800
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.464944444444445e-05,
+      "loss": 0.0047,
+      "step": 57810
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.4648518518518524e-05,
+      "loss": 0.0018,
+      "step": 57820
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.46475925925926e-05,
+      "loss": 0.0035,
+      "step": 57830
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.464666666666667e-05,
+      "loss": 0.0032,
+      "step": 57840
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.464574074074074e-05,
+      "loss": 0.0026,
+      "step": 57850
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.464481481481482e-05,
+      "loss": 0.0025,
+      "step": 57860
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.464388888888889e-05,
+      "loss": 0.0058,
+      "step": 57870
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.464296296296297e-05,
+      "loss": 0.006,
+      "step": 57880
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.4642037037037036e-05,
+      "loss": 0.0022,
+      "step": 57890
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.464111111111111e-05,
+      "loss": 0.0069,
+      "step": 57900
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.4640185185185186e-05,
+      "loss": 0.0026,
+      "step": 57910
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.463925925925926e-05,
+      "loss": 0.0046,
+      "step": 57920
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.4638333333333336e-05,
+      "loss": 0.0031,
+      "step": 57930
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.463740740740741e-05,
+      "loss": 0.0033,
+      "step": 57940
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.4636481481481486e-05,
+      "loss": 0.0034,
+      "step": 57950
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.463555555555556e-05,
+      "loss": 0.0048,
+      "step": 57960
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.463462962962963e-05,
+      "loss": 0.0036,
+      "step": 57970
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.4633703703703704e-05,
+      "loss": 0.0023,
+      "step": 57980
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.463277777777778e-05,
+      "loss": 0.0054,
+      "step": 57990
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.4631851851851855e-05,
+      "loss": 0.0045,
+      "step": 58000
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.463092592592592e-05,
+      "loss": 0.0033,
+      "step": 58010
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.463e-05,
+      "loss": 0.0043,
+      "step": 58020
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.462907407407408e-05,
+      "loss": 0.0031,
+      "step": 58030
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.4628148148148155e-05,
+      "loss": 0.0058,
+      "step": 58040
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.462722222222222e-05,
+      "loss": 0.0061,
+      "step": 58050
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.46262962962963e-05,
+      "loss": 0.0046,
+      "step": 58060
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.462537037037037e-05,
+      "loss": 0.0039,
+      "step": 58070
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.462444444444445e-05,
+      "loss": 0.0023,
+      "step": 58080
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.462351851851852e-05,
+      "loss": 0.0033,
+      "step": 58090
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.462259259259259e-05,
+      "loss": 0.0051,
+      "step": 58100
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.462166666666667e-05,
+      "loss": 0.0021,
+      "step": 58110
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.462074074074074e-05,
+      "loss": 0.0022,
+      "step": 58120
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.461981481481482e-05,
+      "loss": 0.0049,
+      "step": 58130
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.461888888888889e-05,
+      "loss": 0.0077,
+      "step": 58140
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.461796296296297e-05,
+      "loss": 0.004,
+      "step": 58150
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.461703703703704e-05,
+      "loss": 0.0056,
+      "step": 58160
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.461611111111112e-05,
+      "loss": 0.0038,
+      "step": 58170
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.4615185185185185e-05,
+      "loss": 0.003,
+      "step": 58180
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.461425925925926e-05,
+      "loss": 0.0046,
+      "step": 58190
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.4613333333333335e-05,
+      "loss": 0.0035,
+      "step": 58200
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.461240740740741e-05,
+      "loss": 0.0065,
+      "step": 58210
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.461148148148148e-05,
+      "loss": 0.0025,
+      "step": 58220
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.4610555555555554e-05,
+      "loss": 0.0028,
+      "step": 58230
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.460962962962963e-05,
+      "loss": 0.0029,
+      "step": 58240
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.460870370370371e-05,
+      "loss": 0.0047,
+      "step": 58250
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.460777777777778e-05,
+      "loss": 0.0032,
+      "step": 58260
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.4606851851851854e-05,
+      "loss": 0.0036,
+      "step": 58270
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.460592592592593e-05,
+      "loss": 0.006,
+      "step": 58280
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.4605000000000004e-05,
+      "loss": 0.0046,
+      "step": 58290
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.460407407407408e-05,
+      "loss": 0.0022,
+      "step": 58300
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.460314814814815e-05,
+      "loss": 0.0046,
+      "step": 58310
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.460222222222222e-05,
+      "loss": 0.0024,
+      "step": 58320
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.46012962962963e-05,
+      "loss": 0.0033,
+      "step": 58330
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.460037037037037e-05,
+      "loss": 0.0024,
+      "step": 58340
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.459944444444445e-05,
+      "loss": 0.002,
+      "step": 58350
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.459851851851852e-05,
+      "loss": 0.0051,
+      "step": 58360
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.45975925925926e-05,
+      "loss": 0.0039,
+      "step": 58370
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.459666666666667e-05,
+      "loss": 0.0037,
+      "step": 58380
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.459574074074074e-05,
+      "loss": 0.0037,
+      "step": 58390
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.4594814814814816e-05,
+      "loss": 0.0037,
+      "step": 58400
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.459388888888889e-05,
+      "loss": 0.0033,
+      "step": 58410
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.4592962962962966e-05,
+      "loss": 0.0037,
+      "step": 58420
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.4592037037037035e-05,
+      "loss": 0.0059,
+      "step": 58430
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.459111111111111e-05,
+      "loss": 0.0057,
+      "step": 58440
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.4590185185185185e-05,
+      "loss": 0.0018,
+      "step": 58450
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.4589259259259266e-05,
+      "loss": 0.0055,
+      "step": 58460
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.4588333333333335e-05,
+      "loss": 0.0031,
+      "step": 58470
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.458740740740741e-05,
+      "loss": 0.0047,
+      "step": 58480
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.4586481481481485e-05,
+      "loss": 0.0026,
+      "step": 58490
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.458555555555556e-05,
+      "loss": 0.0029,
+      "step": 58500
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.4584629629629635e-05,
+      "loss": 0.0041,
+      "step": 58510
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.45837037037037e-05,
+      "loss": 0.0053,
+      "step": 58520
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.458277777777778e-05,
+      "loss": 0.0017,
+      "step": 58530
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.458185185185185e-05,
+      "loss": 0.0038,
+      "step": 58540
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.458092592592593e-05,
+      "loss": 0.0056,
+      "step": 58550
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.458e-05,
+      "loss": 0.0077,
+      "step": 58560
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.457907407407408e-05,
+      "loss": 0.0036,
+      "step": 58570
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.4578148148148154e-05,
+      "loss": 0.0043,
+      "step": 58580
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.457722222222223e-05,
+      "loss": 0.0027,
+      "step": 58590
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.45762962962963e-05,
+      "loss": 0.004,
+      "step": 58600
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.457537037037037e-05,
+      "loss": 0.0032,
+      "step": 58610
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.457444444444445e-05,
+      "loss": 0.0047,
+      "step": 58620
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.457351851851852e-05,
+      "loss": 0.0061,
+      "step": 58630
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.457259259259259e-05,
+      "loss": 0.0048,
+      "step": 58640
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.4571666666666665e-05,
+      "loss": 0.0027,
+      "step": 58650
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.457074074074074e-05,
+      "loss": 0.0025,
+      "step": 58660
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.4569814814814816e-05,
+      "loss": 0.0035,
+      "step": 58670
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.456888888888889e-05,
+      "loss": 0.0034,
+      "step": 58680
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.4567962962962966e-05,
+      "loss": 0.0026,
+      "step": 58690
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.456703703703704e-05,
+      "loss": 0.0048,
+      "step": 58700
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.4566111111111116e-05,
+      "loss": 0.0031,
+      "step": 58710
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.456518518518519e-05,
+      "loss": 0.0023,
+      "step": 58720
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.456425925925926e-05,
+      "loss": 0.0053,
+      "step": 58730
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.4563333333333334e-05,
+      "loss": 0.0029,
+      "step": 58740
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.456240740740741e-05,
+      "loss": 0.002,
+      "step": 58750
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.4561481481481484e-05,
+      "loss": 0.0044,
+      "step": 58760
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.456055555555555e-05,
+      "loss": 0.0034,
+      "step": 58770
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.4559629629629634e-05,
+      "loss": 0.0017,
+      "step": 58780
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.455870370370371e-05,
+      "loss": 0.005,
+      "step": 58790
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.4557777777777784e-05,
+      "loss": 0.0039,
+      "step": 58800
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.455685185185185e-05,
+      "loss": 0.0027,
+      "step": 58810
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.455592592592593e-05,
+      "loss": 0.0065,
+      "step": 58820
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.4555e-05,
+      "loss": 0.0032,
+      "step": 58830
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.455416666666667e-05,
+      "loss": 0.0046,
+      "step": 58840
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.455324074074074e-05,
+      "loss": 0.0035,
+      "step": 58850
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.455231481481482e-05,
+      "loss": 0.004,
+      "step": 58860
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.455138888888889e-05,
+      "loss": 0.0052,
+      "step": 58870
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.455046296296296e-05,
+      "loss": 0.004,
+      "step": 58880
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.4549537037037036e-05,
+      "loss": 0.0023,
+      "step": 58890
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.454861111111112e-05,
+      "loss": 0.0047,
+      "step": 58900
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.4547685185185186e-05,
+      "loss": 0.004,
+      "step": 58910
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.454675925925926e-05,
+      "loss": 0.0023,
+      "step": 58920
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.4545833333333336e-05,
+      "loss": 0.0039,
+      "step": 58930
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.454490740740741e-05,
+      "loss": 0.0031,
+      "step": 58940
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.4543981481481487e-05,
+      "loss": 0.0037,
+      "step": 58950
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.4543055555555555e-05,
+      "loss": 0.0031,
+      "step": 58960
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.454212962962963e-05,
+      "loss": 0.002,
+      "step": 58970
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.4541203703703705e-05,
+      "loss": 0.0037,
+      "step": 58980
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.454027777777778e-05,
+      "loss": 0.0033,
+      "step": 58990
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.453935185185185e-05,
+      "loss": 0.0069,
+      "step": 59000
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.453842592592593e-05,
+      "loss": 0.0042,
+      "step": 59010
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.4537500000000005e-05,
+      "loss": 0.0036,
+      "step": 59020
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.453657407407408e-05,
+      "loss": 0.0046,
+      "step": 59030
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.453564814814815e-05,
+      "loss": 0.0054,
+      "step": 59040
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.4534722222222224e-05,
+      "loss": 0.003,
+      "step": 59050
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.45337962962963e-05,
+      "loss": 0.0033,
+      "step": 59060
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.4532870370370374e-05,
+      "loss": 0.0047,
+      "step": 59070
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.453194444444445e-05,
+      "loss": 0.0032,
+      "step": 59080
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.453101851851852e-05,
+      "loss": 0.0041,
+      "step": 59090
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.453009259259259e-05,
+      "loss": 0.0051,
+      "step": 59100
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.452916666666667e-05,
+      "loss": 0.0024,
+      "step": 59110
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.452824074074074e-05,
+      "loss": 0.0041,
+      "step": 59120
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.452731481481482e-05,
+      "loss": 0.0057,
+      "step": 59130
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.452638888888889e-05,
+      "loss": 0.0059,
+      "step": 59140
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.452546296296297e-05,
+      "loss": 0.0037,
+      "step": 59150
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.452453703703704e-05,
+      "loss": 0.0047,
+      "step": 59160
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.452361111111111e-05,
+      "loss": 0.0029,
+      "step": 59170
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.4522685185185186e-05,
+      "loss": 0.004,
+      "step": 59180
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.452175925925926e-05,
+      "loss": 0.0034,
+      "step": 59190
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.4520833333333336e-05,
+      "loss": 0.0018,
+      "step": 59200
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.4519907407407404e-05,
+      "loss": 0.0036,
+      "step": 59210
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.451898148148148e-05,
+      "loss": 0.0034,
+      "step": 59220
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.451805555555556e-05,
+      "loss": 0.0014,
+      "step": 59230
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.4517129629629636e-05,
+      "loss": 0.0027,
+      "step": 59240
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.4516203703703704e-05,
+      "loss": 0.0022,
+      "step": 59250
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.451527777777778e-05,
+      "loss": 0.0032,
+      "step": 59260
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.4514351851851854e-05,
+      "loss": 0.008,
+      "step": 59270
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.451342592592593e-05,
+      "loss": 0.0028,
+      "step": 59280
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.4512500000000005e-05,
+      "loss": 0.0036,
+      "step": 59290
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.451157407407407e-05,
+      "loss": 0.0035,
+      "step": 59300
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.451064814814815e-05,
+      "loss": 0.0033,
+      "step": 59310
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.450972222222222e-05,
+      "loss": 0.0037,
+      "step": 59320
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.45087962962963e-05,
+      "loss": 0.0036,
+      "step": 59330
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.450787037037037e-05,
+      "loss": 0.006,
+      "step": 59340
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.450694444444445e-05,
+      "loss": 0.0033,
+      "step": 59350
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.450601851851852e-05,
+      "loss": 0.007,
+      "step": 59360
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.45050925925926e-05,
+      "loss": 0.0027,
+      "step": 59370
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.4504166666666666e-05,
+      "loss": 0.0046,
+      "step": 59380
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.450324074074074e-05,
+      "loss": 0.0052,
+      "step": 59390
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.4502314814814817e-05,
+      "loss": 0.005,
+      "step": 59400
+    },
+    {
+      "epoch": 11.0,
+      "eval_loss": 0.013140509836375713,
+      "eval_runtime": 153.62,
+      "eval_samples_per_second": 25.973,
+      "eval_steps_per_second": 3.248,
+      "step": 59400
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.450138888888889e-05,
+      "loss": 0.004,
+      "step": 59410
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.450046296296296e-05,
+      "loss": 0.0032,
+      "step": 59420
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.4499537037037035e-05,
+      "loss": 0.0032,
+      "step": 59430
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.449861111111112e-05,
+      "loss": 0.0026,
+      "step": 59440
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.449768518518519e-05,
+      "loss": 0.004,
+      "step": 59450
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.449675925925926e-05,
+      "loss": 0.0031,
+      "step": 59460
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.4495833333333335e-05,
+      "loss": 0.0029,
+      "step": 59470
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.449490740740741e-05,
+      "loss": 0.0043,
+      "step": 59480
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.4493981481481485e-05,
+      "loss": 0.0017,
+      "step": 59490
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.449305555555556e-05,
+      "loss": 0.0029,
+      "step": 59500
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.449212962962963e-05,
+      "loss": 0.0026,
+      "step": 59510
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.4491203703703704e-05,
+      "loss": 0.0039,
+      "step": 59520
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.449027777777778e-05,
+      "loss": 0.0017,
+      "step": 59530
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.4489351851851854e-05,
+      "loss": 0.0024,
+      "step": 59540
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.448842592592593e-05,
+      "loss": 0.0037,
+      "step": 59550
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.4487500000000004e-05,
+      "loss": 0.0033,
+      "step": 59560
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.448657407407408e-05,
+      "loss": 0.0031,
+      "step": 59570
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.4485648148148154e-05,
+      "loss": 0.0047,
+      "step": 59580
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.448472222222222e-05,
+      "loss": 0.0027,
+      "step": 59590
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.44837962962963e-05,
+      "loss": 0.0029,
+      "step": 59600
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.448287037037037e-05,
+      "loss": 0.0022,
+      "step": 59610
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.448194444444445e-05,
+      "loss": 0.0024,
+      "step": 59620
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.4481018518518516e-05,
+      "loss": 0.0029,
+      "step": 59630
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.448009259259259e-05,
+      "loss": 0.0042,
+      "step": 59640
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.4479166666666666e-05,
+      "loss": 0.0038,
+      "step": 59650
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.447824074074075e-05,
+      "loss": 0.0029,
+      "step": 59660
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.4477314814814816e-05,
+      "loss": 0.0039,
+      "step": 59670
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.447638888888889e-05,
+      "loss": 0.0035,
+      "step": 59680
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.4475462962962966e-05,
+      "loss": 0.0075,
+      "step": 59690
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.447453703703704e-05,
+      "loss": 0.0026,
+      "step": 59700
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.4473611111111116e-05,
+      "loss": 0.0034,
+      "step": 59710
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.4472685185185184e-05,
+      "loss": 0.0021,
+      "step": 59720
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.447175925925926e-05,
+      "loss": 0.0029,
+      "step": 59730
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.4470833333333335e-05,
+      "loss": 0.0032,
+      "step": 59740
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.446990740740741e-05,
+      "loss": 0.0026,
+      "step": 59750
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.446898148148148e-05,
+      "loss": 0.0031,
+      "step": 59760
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.446805555555556e-05,
+      "loss": 0.0031,
+      "step": 59770
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.4467129629629635e-05,
+      "loss": 0.0023,
+      "step": 59780
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.446620370370371e-05,
+      "loss": 0.0043,
+      "step": 59790
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.446527777777778e-05,
+      "loss": 0.003,
+      "step": 59800
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.446435185185185e-05,
+      "loss": 0.0022,
+      "step": 59810
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.446342592592593e-05,
+      "loss": 0.0036,
+      "step": 59820
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.44625e-05,
+      "loss": 0.0019,
+      "step": 59830
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.446157407407407e-05,
+      "loss": 0.0029,
+      "step": 59840
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.446064814814815e-05,
+      "loss": 0.0041,
+      "step": 59850
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.445972222222222e-05,
+      "loss": 0.0047,
+      "step": 59860
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.4458796296296304e-05,
+      "loss": 0.0032,
+      "step": 59870
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.445787037037037e-05,
+      "loss": 0.002,
+      "step": 59880
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.445694444444445e-05,
+      "loss": 0.0073,
+      "step": 59890
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.445601851851852e-05,
+      "loss": 0.0034,
+      "step": 59900
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.44550925925926e-05,
+      "loss": 0.0026,
+      "step": 59910
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.445416666666667e-05,
+      "loss": 0.003,
+      "step": 59920
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.445324074074074e-05,
+      "loss": 0.0037,
+      "step": 59930
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.4452314814814815e-05,
+      "loss": 0.0071,
+      "step": 59940
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.445138888888889e-05,
+      "loss": 0.0045,
+      "step": 59950
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.4450462962962965e-05,
+      "loss": 0.003,
+      "step": 59960
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.4449537037037034e-05,
+      "loss": 0.0063,
+      "step": 59970
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.4448611111111116e-05,
+      "loss": 0.002,
+      "step": 59980
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.444768518518519e-05,
+      "loss": 0.0043,
+      "step": 59990
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.4446759259259266e-05,
+      "loss": 0.0052,
+      "step": 60000
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.4445833333333334e-05,
+      "loss": 0.0032,
+      "step": 60010
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.444490740740741e-05,
+      "loss": 0.0023,
+      "step": 60020
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.4443981481481484e-05,
+      "loss": 0.0026,
+      "step": 60030
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.444305555555556e-05,
+      "loss": 0.0042,
+      "step": 60040
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.444212962962963e-05,
+      "loss": 0.004,
+      "step": 60050
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.44412037037037e-05,
+      "loss": 0.0024,
+      "step": 60060
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.444027777777778e-05,
+      "loss": 0.0042,
+      "step": 60070
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.443935185185185e-05,
+      "loss": 0.0049,
+      "step": 60080
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.443842592592593e-05,
+      "loss": 0.0044,
+      "step": 60090
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.44375e-05,
+      "loss": 0.0063,
+      "step": 60100
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.443657407407408e-05,
+      "loss": 0.0028,
+      "step": 60110
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.443564814814815e-05,
+      "loss": 0.0049,
+      "step": 60120
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.443472222222223e-05,
+      "loss": 0.0024,
+      "step": 60130
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.4433796296296296e-05,
+      "loss": 0.0049,
+      "step": 60140
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.443287037037037e-05,
+      "loss": 0.0032,
+      "step": 60150
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.4431944444444446e-05,
+      "loss": 0.0023,
+      "step": 60160
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.443101851851852e-05,
+      "loss": 0.0038,
+      "step": 60170
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.443009259259259e-05,
+      "loss": 0.0033,
+      "step": 60180
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.4429166666666665e-05,
+      "loss": 0.0024,
+      "step": 60190
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.4428240740740746e-05,
+      "loss": 0.0055,
+      "step": 60200
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.442731481481482e-05,
+      "loss": 0.0025,
+      "step": 60210
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.442638888888889e-05,
+      "loss": 0.0046,
+      "step": 60220
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.4425462962962965e-05,
+      "loss": 0.0044,
+      "step": 60230
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.442453703703704e-05,
+      "loss": 0.0037,
+      "step": 60240
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.4423611111111115e-05,
+      "loss": 0.0026,
+      "step": 60250
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.442268518518518e-05,
+      "loss": 0.0029,
+      "step": 60260
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.442175925925926e-05,
+      "loss": 0.0029,
+      "step": 60270
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.442083333333333e-05,
+      "loss": 0.0053,
+      "step": 60280
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.441990740740741e-05,
+      "loss": 0.0026,
+      "step": 60290
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.4418981481481483e-05,
+      "loss": 0.0027,
+      "step": 60300
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.441805555555556e-05,
+      "loss": 0.0048,
+      "step": 60310
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.4417129629629634e-05,
+      "loss": 0.0032,
+      "step": 60320
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.441620370370371e-05,
+      "loss": 0.0016,
+      "step": 60330
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.4415277777777784e-05,
+      "loss": 0.0033,
+      "step": 60340
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.441435185185185e-05,
+      "loss": 0.0044,
+      "step": 60350
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.441342592592593e-05,
+      "loss": 0.0063,
+      "step": 60360
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.44125e-05,
+      "loss": 0.0024,
+      "step": 60370
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.441157407407408e-05,
+      "loss": 0.0033,
+      "step": 60380
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.4410648148148145e-05,
+      "loss": 0.003,
+      "step": 60390
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.440972222222222e-05,
+      "loss": 0.0027,
+      "step": 60400
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.44087962962963e-05,
+      "loss": 0.0033,
+      "step": 60410
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.440787037037038e-05,
+      "loss": 0.0036,
+      "step": 60420
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.4406944444444446e-05,
+      "loss": 0.0039,
+      "step": 60430
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.440601851851852e-05,
+      "loss": 0.0041,
+      "step": 60440
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.4405092592592596e-05,
+      "loss": 0.0023,
+      "step": 60450
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.440416666666667e-05,
+      "loss": 0.0022,
+      "step": 60460
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.440324074074074e-05,
+      "loss": 0.0029,
+      "step": 60470
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.4402314814814814e-05,
+      "loss": 0.0027,
+      "step": 60480
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.440138888888889e-05,
+      "loss": 0.0022,
+      "step": 60490
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.4400462962962964e-05,
+      "loss": 0.0037,
+      "step": 60500
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.439953703703704e-05,
+      "loss": 0.0058,
+      "step": 60510
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.4398611111111114e-05,
+      "loss": 0.0039,
+      "step": 60520
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.439768518518519e-05,
+      "loss": 0.0023,
+      "step": 60530
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.4396759259259264e-05,
+      "loss": 0.0028,
+      "step": 60540
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.439583333333334e-05,
+      "loss": 0.0035,
+      "step": 60550
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.439490740740741e-05,
+      "loss": 0.0034,
+      "step": 60560
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.439398148148148e-05,
+      "loss": 0.0039,
+      "step": 60570
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.439305555555556e-05,
+      "loss": 0.0027,
+      "step": 60580
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.439212962962963e-05,
+      "loss": 0.0038,
+      "step": 60590
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.43912037037037e-05,
+      "loss": 0.0017,
+      "step": 60600
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.4390277777777776e-05,
+      "loss": 0.0037,
+      "step": 60610
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.438935185185185e-05,
+      "loss": 0.0032,
+      "step": 60620
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.438842592592593e-05,
+      "loss": 0.0039,
+      "step": 60630
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.43875e-05,
+      "loss": 0.0032,
+      "step": 60640
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.4386574074074077e-05,
+      "loss": 0.0042,
+      "step": 60650
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.438564814814815e-05,
+      "loss": 0.0044,
+      "step": 60660
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.438472222222223e-05,
+      "loss": 0.0029,
+      "step": 60670
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.4383796296296295e-05,
+      "loss": 0.0046,
+      "step": 60680
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.438287037037037e-05,
+      "loss": 0.0035,
+      "step": 60690
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.4381944444444445e-05,
+      "loss": 0.0026,
+      "step": 60700
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.438101851851852e-05,
+      "loss": 0.0036,
+      "step": 60710
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.4380092592592595e-05,
+      "loss": 0.0047,
+      "step": 60720
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.4379166666666663e-05,
+      "loss": 0.0038,
+      "step": 60730
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.4378240740740745e-05,
+      "loss": 0.0029,
+      "step": 60740
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.437731481481482e-05,
+      "loss": 0.0046,
+      "step": 60750
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.4376388888888895e-05,
+      "loss": 0.0048,
+      "step": 60760
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.4375462962962964e-05,
+      "loss": 0.0028,
+      "step": 60770
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.437453703703704e-05,
+      "loss": 0.0055,
+      "step": 60780
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.4373611111111114e-05,
+      "loss": 0.0036,
+      "step": 60790
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.437268518518519e-05,
+      "loss": 0.0068,
+      "step": 60800
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.437175925925926e-05,
+      "loss": 0.0034,
+      "step": 60810
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.437083333333333e-05,
+      "loss": 0.0028,
+      "step": 60820
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.436990740740741e-05,
+      "loss": 0.0036,
+      "step": 60830
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.436898148148149e-05,
+      "loss": 0.0032,
+      "step": 60840
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.436805555555556e-05,
+      "loss": 0.0028,
+      "step": 60850
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.436712962962963e-05,
+      "loss": 0.0039,
+      "step": 60860
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.436620370370371e-05,
+      "loss": 0.0029,
+      "step": 60870
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.436527777777778e-05,
+      "loss": 0.0028,
+      "step": 60880
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.436435185185185e-05,
+      "loss": 0.0025,
+      "step": 60890
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.4363425925925926e-05,
+      "loss": 0.0042,
+      "step": 60900
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.43625e-05,
+      "loss": 0.0026,
+      "step": 60910
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.4361574074074076e-05,
+      "loss": 0.0034,
+      "step": 60920
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.436064814814815e-05,
+      "loss": 0.0027,
+      "step": 60930
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.435972222222222e-05,
+      "loss": 0.0029,
+      "step": 60940
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.43587962962963e-05,
+      "loss": 0.0041,
+      "step": 60950
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.4357870370370376e-05,
+      "loss": 0.003,
+      "step": 60960
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.435694444444445e-05,
+      "loss": 0.0038,
+      "step": 60970
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.435601851851852e-05,
+      "loss": 0.0044,
+      "step": 60980
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.4355092592592595e-05,
+      "loss": 0.005,
+      "step": 60990
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.435416666666667e-05,
+      "loss": 0.0052,
+      "step": 61000
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.4353240740740745e-05,
+      "loss": 0.0038,
+      "step": 61010
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.435231481481481e-05,
+      "loss": 0.0052,
+      "step": 61020
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.435138888888889e-05,
+      "loss": 0.0026,
+      "step": 61030
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.435046296296296e-05,
+      "loss": 0.0047,
+      "step": 61040
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.434953703703704e-05,
+      "loss": 0.0041,
+      "step": 61050
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.434861111111111e-05,
+      "loss": 0.0066,
+      "step": 61060
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.434768518518519e-05,
+      "loss": 0.0047,
+      "step": 61070
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.434675925925926e-05,
+      "loss": 0.0037,
+      "step": 61080
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.434583333333334e-05,
+      "loss": 0.0027,
+      "step": 61090
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.4344907407407413e-05,
+      "loss": 0.0037,
+      "step": 61100
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.434398148148148e-05,
+      "loss": 0.0039,
+      "step": 61110
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.434305555555556e-05,
+      "loss": 0.0023,
+      "step": 61120
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.434212962962963e-05,
+      "loss": 0.0027,
+      "step": 61130
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.434120370370371e-05,
+      "loss": 0.0055,
+      "step": 61140
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.4340277777777775e-05,
+      "loss": 0.0032,
+      "step": 61150
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.433944444444445e-05,
+      "loss": 0.0047,
+      "step": 61160
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.4338518518518515e-05,
+      "loss": 0.002,
+      "step": 61170
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.43375925925926e-05,
+      "loss": 0.003,
+      "step": 61180
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.433666666666667e-05,
+      "loss": 0.0027,
+      "step": 61190
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.433574074074075e-05,
+      "loss": 0.0048,
+      "step": 61200
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.4334814814814815e-05,
+      "loss": 0.0052,
+      "step": 61210
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.433388888888889e-05,
+      "loss": 0.0044,
+      "step": 61220
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.4332962962962965e-05,
+      "loss": 0.0054,
+      "step": 61230
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.433203703703704e-05,
+      "loss": 0.0069,
+      "step": 61240
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.4331111111111115e-05,
+      "loss": 0.006,
+      "step": 61250
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.4330185185185184e-05,
+      "loss": 0.0033,
+      "step": 61260
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.432925925925926e-05,
+      "loss": 0.002,
+      "step": 61270
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.4328333333333334e-05,
+      "loss": 0.0029,
+      "step": 61280
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.432740740740741e-05,
+      "loss": 0.0045,
+      "step": 61290
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.4326481481481484e-05,
+      "loss": 0.003,
+      "step": 61300
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.432555555555556e-05,
+      "loss": 0.0018,
+      "step": 61310
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.4324629629629634e-05,
+      "loss": 0.0035,
+      "step": 61320
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.432370370370371e-05,
+      "loss": 0.0035,
+      "step": 61330
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.432277777777778e-05,
+      "loss": 0.0042,
+      "step": 61340
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.432185185185185e-05,
+      "loss": 0.0033,
+      "step": 61350
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.432092592592593e-05,
+      "loss": 0.0026,
+      "step": 61360
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.432e-05,
+      "loss": 0.003,
+      "step": 61370
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.431907407407407e-05,
+      "loss": 0.0027,
+      "step": 61380
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.4318148148148146e-05,
+      "loss": 0.0016,
+      "step": 61390
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.431722222222223e-05,
+      "loss": 0.0026,
+      "step": 61400
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.43162962962963e-05,
+      "loss": 0.0054,
+      "step": 61410
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.431537037037037e-05,
+      "loss": 0.0032,
+      "step": 61420
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.4314444444444446e-05,
+      "loss": 0.0024,
+      "step": 61430
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.431351851851852e-05,
+      "loss": 0.007,
+      "step": 61440
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.4312592592592596e-05,
+      "loss": 0.0029,
+      "step": 61450
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.431166666666667e-05,
+      "loss": 0.0029,
+      "step": 61460
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.431074074074074e-05,
+      "loss": 0.0049,
+      "step": 61470
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.4309814814814815e-05,
+      "loss": 0.0025,
+      "step": 61480
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.430888888888889e-05,
+      "loss": 0.0047,
+      "step": 61490
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.4307962962962965e-05,
+      "loss": 0.0021,
+      "step": 61500
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.430703703703704e-05,
+      "loss": 0.0057,
+      "step": 61510
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.4306111111111115e-05,
+      "loss": 0.0034,
+      "step": 61520
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.430518518518519e-05,
+      "loss": 0.0033,
+      "step": 61530
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.4304259259259265e-05,
+      "loss": 0.0037,
+      "step": 61540
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.430333333333333e-05,
+      "loss": 0.0051,
+      "step": 61550
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.430240740740741e-05,
+      "loss": 0.0081,
+      "step": 61560
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.430148148148148e-05,
+      "loss": 0.0034,
+      "step": 61570
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.430055555555556e-05,
+      "loss": 0.0041,
+      "step": 61580
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.429962962962963e-05,
+      "loss": 0.0022,
+      "step": 61590
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.42987037037037e-05,
+      "loss": 0.004,
+      "step": 61600
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.4297777777777784e-05,
+      "loss": 0.004,
+      "step": 61610
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.429685185185186e-05,
+      "loss": 0.0068,
+      "step": 61620
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.429592592592593e-05,
+      "loss": 0.0029,
+      "step": 61630
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.4295e-05,
+      "loss": 0.0049,
+      "step": 61640
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.429407407407408e-05,
+      "loss": 0.0026,
+      "step": 61650
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.429314814814815e-05,
+      "loss": 0.0082,
+      "step": 61660
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.429222222222223e-05,
+      "loss": 0.0054,
+      "step": 61670
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.4291296296296295e-05,
+      "loss": 0.0047,
+      "step": 61680
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.429037037037037e-05,
+      "loss": 0.0033,
+      "step": 61690
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.4289444444444446e-05,
+      "loss": 0.003,
+      "step": 61700
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.428851851851852e-05,
+      "loss": 0.0017,
+      "step": 61710
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.4287592592592596e-05,
+      "loss": 0.0033,
+      "step": 61720
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.428666666666667e-05,
+      "loss": 0.0051,
+      "step": 61730
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.4285740740740746e-05,
+      "loss": 0.0016,
+      "step": 61740
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.428481481481482e-05,
+      "loss": 0.0035,
+      "step": 61750
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.428388888888889e-05,
+      "loss": 0.0047,
+      "step": 61760
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.4282962962962964e-05,
+      "loss": 0.0039,
+      "step": 61770
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.428203703703704e-05,
+      "loss": 0.0048,
+      "step": 61780
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.4281111111111114e-05,
+      "loss": 0.0041,
+      "step": 61790
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.428018518518518e-05,
+      "loss": 0.003,
+      "step": 61800
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.427925925925926e-05,
+      "loss": 0.003,
+      "step": 61810
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.427833333333333e-05,
+      "loss": 0.0035,
+      "step": 61820
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.4277407407407414e-05,
+      "loss": 0.0045,
+      "step": 61830
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.427648148148148e-05,
+      "loss": 0.0042,
+      "step": 61840
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.427555555555556e-05,
+      "loss": 0.0041,
+      "step": 61850
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.427462962962963e-05,
+      "loss": 0.0028,
+      "step": 61860
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.427370370370371e-05,
+      "loss": 0.0026,
+      "step": 61870
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.427277777777778e-05,
+      "loss": 0.0038,
+      "step": 61880
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.427185185185185e-05,
+      "loss": 0.0036,
+      "step": 61890
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.4270925925925926e-05,
+      "loss": 0.0056,
+      "step": 61900
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.427e-05,
+      "loss": 0.0026,
+      "step": 61910
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.4269074074074076e-05,
+      "loss": 0.0065,
+      "step": 61920
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.426814814814815e-05,
+      "loss": 0.0036,
+      "step": 61930
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.4267222222222227e-05,
+      "loss": 0.0029,
+      "step": 61940
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.42662962962963e-05,
+      "loss": 0.0053,
+      "step": 61950
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.426537037037038e-05,
+      "loss": 0.0048,
+      "step": 61960
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.4264444444444445e-05,
+      "loss": 0.0026,
+      "step": 61970
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.426351851851852e-05,
+      "loss": 0.0034,
+      "step": 61980
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.4262592592592595e-05,
+      "loss": 0.0027,
+      "step": 61990
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.426166666666667e-05,
+      "loss": 0.0026,
+      "step": 62000
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.426074074074074e-05,
+      "loss": 0.0038,
+      "step": 62010
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.4259814814814813e-05,
+      "loss": 0.0045,
+      "step": 62020
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.425888888888889e-05,
+      "loss": 0.0063,
+      "step": 62030
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.425796296296297e-05,
+      "loss": 0.0061,
+      "step": 62040
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.425703703703704e-05,
+      "loss": 0.0022,
+      "step": 62050
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.4256111111111114e-05,
+      "loss": 0.0045,
+      "step": 62060
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.425518518518519e-05,
+      "loss": 0.0032,
+      "step": 62070
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.4254259259259264e-05,
+      "loss": 0.0033,
+      "step": 62080
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.425333333333334e-05,
+      "loss": 0.0034,
+      "step": 62090
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.425240740740741e-05,
+      "loss": 0.0027,
+      "step": 62100
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.425148148148148e-05,
+      "loss": 0.0033,
+      "step": 62110
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.425055555555556e-05,
+      "loss": 0.0046,
+      "step": 62120
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.424962962962963e-05,
+      "loss": 0.0023,
+      "step": 62130
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.42487037037037e-05,
+      "loss": 0.0044,
+      "step": 62140
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.424777777777778e-05,
+      "loss": 0.005,
+      "step": 62150
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.424685185185186e-05,
+      "loss": 0.0039,
+      "step": 62160
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.424592592592593e-05,
+      "loss": 0.0045,
+      "step": 62170
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.4245e-05,
+      "loss": 0.0025,
+      "step": 62180
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.4244074074074076e-05,
+      "loss": 0.0052,
+      "step": 62190
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.424314814814815e-05,
+      "loss": 0.0076,
+      "step": 62200
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.4242222222222226e-05,
+      "loss": 0.0038,
+      "step": 62210
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.4241296296296294e-05,
+      "loss": 0.0032,
+      "step": 62220
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.424037037037037e-05,
+      "loss": 0.0041,
+      "step": 62230
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.4239444444444444e-05,
+      "loss": 0.0038,
+      "step": 62240
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.423851851851852e-05,
+      "loss": 0.003,
+      "step": 62250
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.4237592592592594e-05,
+      "loss": 0.0034,
+      "step": 62260
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.423666666666667e-05,
+      "loss": 0.0042,
+      "step": 62270
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.4235740740740745e-05,
+      "loss": 0.0036,
+      "step": 62280
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.423481481481482e-05,
+      "loss": 0.0028,
+      "step": 62290
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.4233888888888895e-05,
+      "loss": 0.0027,
+      "step": 62300
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.423296296296296e-05,
+      "loss": 0.0041,
+      "step": 62310
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.423203703703704e-05,
+      "loss": 0.0027,
+      "step": 62320
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.423111111111111e-05,
+      "loss": 0.0031,
+      "step": 62330
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.423018518518519e-05,
+      "loss": 0.0052,
+      "step": 62340
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.4229259259259256e-05,
+      "loss": 0.0047,
+      "step": 62350
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.422833333333333e-05,
+      "loss": 0.005,
+      "step": 62360
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.422740740740741e-05,
+      "loss": 0.0036,
+      "step": 62370
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.422648148148149e-05,
+      "loss": 0.0043,
+      "step": 62380
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.4225555555555557e-05,
+      "loss": 0.0056,
+      "step": 62390
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.422462962962963e-05,
+      "loss": 0.0061,
+      "step": 62400
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.422370370370371e-05,
+      "loss": 0.004,
+      "step": 62410
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.422277777777778e-05,
+      "loss": 0.0049,
+      "step": 62420
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.422185185185185e-05,
+      "loss": 0.0028,
+      "step": 62430
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.4220925925925925e-05,
+      "loss": 0.0026,
+      "step": 62440
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.422e-05,
+      "loss": 0.0042,
+      "step": 62450
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.4219074074074075e-05,
+      "loss": 0.0031,
+      "step": 62460
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.421814814814815e-05,
+      "loss": 0.0041,
+      "step": 62470
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.4217222222222225e-05,
+      "loss": 0.0016,
+      "step": 62480
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.42162962962963e-05,
+      "loss": 0.0036,
+      "step": 62490
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.4215370370370375e-05,
+      "loss": 0.0029,
+      "step": 62500
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.421444444444445e-05,
+      "loss": 0.004,
+      "step": 62510
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.421351851851852e-05,
+      "loss": 0.0059,
+      "step": 62520
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.4212592592592594e-05,
+      "loss": 0.0026,
+      "step": 62530
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.421166666666667e-05,
+      "loss": 0.0027,
+      "step": 62540
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.4210740740740744e-05,
+      "loss": 0.0057,
+      "step": 62550
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.420981481481481e-05,
+      "loss": 0.0043,
+      "step": 62560
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.420888888888889e-05,
+      "loss": 0.0025,
+      "step": 62570
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.420796296296297e-05,
+      "loss": 0.0053,
+      "step": 62580
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.4207037037037044e-05,
+      "loss": 0.0024,
+      "step": 62590
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.420611111111111e-05,
+      "loss": 0.0021,
+      "step": 62600
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.420518518518519e-05,
+      "loss": 0.0038,
+      "step": 62610
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.420425925925926e-05,
+      "loss": 0.0051,
+      "step": 62620
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.420333333333334e-05,
+      "loss": 0.0017,
+      "step": 62630
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.4202407407407406e-05,
+      "loss": 0.0068,
+      "step": 62640
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.420148148148148e-05,
+      "loss": 0.0039,
+      "step": 62650
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.4200555555555556e-05,
+      "loss": 0.0044,
+      "step": 62660
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.419962962962963e-05,
+      "loss": 0.0052,
+      "step": 62670
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.4198703703703706e-05,
+      "loss": 0.0054,
+      "step": 62680
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.419777777777778e-05,
+      "loss": 0.0025,
+      "step": 62690
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.4196851851851856e-05,
+      "loss": 0.0042,
+      "step": 62700
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.419592592592593e-05,
+      "loss": 0.0048,
+      "step": 62710
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.4195000000000006e-05,
+      "loss": 0.0037,
+      "step": 62720
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.4194074074074075e-05,
+      "loss": 0.0026,
+      "step": 62730
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.419314814814815e-05,
+      "loss": 0.0023,
+      "step": 62740
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.4192222222222225e-05,
+      "loss": 0.0029,
+      "step": 62750
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.41912962962963e-05,
+      "loss": 0.0021,
+      "step": 62760
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.419037037037037e-05,
+      "loss": 0.0026,
+      "step": 62770
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.418944444444444e-05,
+      "loss": 0.0033,
+      "step": 62780
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.418851851851852e-05,
+      "loss": 0.0038,
+      "step": 62790
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.41875925925926e-05,
+      "loss": 0.0023,
+      "step": 62800
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.418666666666667e-05,
+      "loss": 0.0027,
+      "step": 62810
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.418574074074074e-05,
+      "loss": 0.0043,
+      "step": 62820
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.418481481481482e-05,
+      "loss": 0.0044,
+      "step": 62830
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.4183888888888893e-05,
+      "loss": 0.0036,
+      "step": 62840
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.418296296296296e-05,
+      "loss": 0.0035,
+      "step": 62850
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.418203703703704e-05,
+      "loss": 0.0029,
+      "step": 62860
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.418111111111111e-05,
+      "loss": 0.0037,
+      "step": 62870
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.418018518518519e-05,
+      "loss": 0.0082,
+      "step": 62880
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.417925925925926e-05,
+      "loss": 0.0033,
+      "step": 62890
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.417833333333333e-05,
+      "loss": 0.0032,
+      "step": 62900
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.417740740740741e-05,
+      "loss": 0.0041,
+      "step": 62910
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.417648148148149e-05,
+      "loss": 0.0049,
+      "step": 62920
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.417555555555556e-05,
+      "loss": 0.0042,
+      "step": 62930
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.417462962962963e-05,
+      "loss": 0.0048,
+      "step": 62940
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.4173703703703705e-05,
+      "loss": 0.0057,
+      "step": 62950
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.417277777777778e-05,
+      "loss": 0.0035,
+      "step": 62960
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.4171851851851856e-05,
+      "loss": 0.003,
+      "step": 62970
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.4170925925925924e-05,
+      "loss": 0.0022,
+      "step": 62980
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.417e-05,
+      "loss": 0.004,
+      "step": 62990
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.4169074074074074e-05,
+      "loss": 0.0019,
+      "step": 63000
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.4168148148148156e-05,
+      "loss": 0.0024,
+      "step": 63010
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.4167222222222224e-05,
+      "loss": 0.0019,
+      "step": 63020
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.41662962962963e-05,
+      "loss": 0.0063,
+      "step": 63030
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.4165370370370374e-05,
+      "loss": 0.0039,
+      "step": 63040
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.416444444444445e-05,
+      "loss": 0.0053,
+      "step": 63050
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.416351851851852e-05,
+      "loss": 0.004,
+      "step": 63060
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.416259259259259e-05,
+      "loss": 0.003,
+      "step": 63070
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.416166666666667e-05,
+      "loss": 0.0048,
+      "step": 63080
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.416074074074074e-05,
+      "loss": 0.002,
+      "step": 63090
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.415981481481482e-05,
+      "loss": 0.0044,
+      "step": 63100
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.4158888888888886e-05,
+      "loss": 0.0026,
+      "step": 63110
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.415796296296297e-05,
+      "loss": 0.0036,
+      "step": 63120
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.415703703703704e-05,
+      "loss": 0.0035,
+      "step": 63130
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.415611111111112e-05,
+      "loss": 0.0029,
+      "step": 63140
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.4155185185185186e-05,
+      "loss": 0.0023,
+      "step": 63150
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.415425925925926e-05,
+      "loss": 0.0034,
+      "step": 63160
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.4153333333333336e-05,
+      "loss": 0.0025,
+      "step": 63170
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.415240740740741e-05,
+      "loss": 0.0033,
+      "step": 63180
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.415148148148148e-05,
+      "loss": 0.0033,
+      "step": 63190
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.4150555555555555e-05,
+      "loss": 0.0025,
+      "step": 63200
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.414962962962963e-05,
+      "loss": 0.0052,
+      "step": 63210
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.4148703703703705e-05,
+      "loss": 0.0054,
+      "step": 63220
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.414777777777778e-05,
+      "loss": 0.0034,
+      "step": 63230
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.4146851851851855e-05,
+      "loss": 0.0023,
+      "step": 63240
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.414592592592593e-05,
+      "loss": 0.0027,
+      "step": 63250
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.4145000000000005e-05,
+      "loss": 0.004,
+      "step": 63260
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.414407407407407e-05,
+      "loss": 0.0029,
+      "step": 63270
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.414314814814815e-05,
+      "loss": 0.0026,
+      "step": 63280
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.4142222222222223e-05,
+      "loss": 0.0032,
+      "step": 63290
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.41412962962963e-05,
+      "loss": 0.0049,
+      "step": 63300
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.4140370370370374e-05,
+      "loss": 0.0055,
+      "step": 63310
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.413944444444444e-05,
+      "loss": 0.0032,
+      "step": 63320
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.413851851851852e-05,
+      "loss": 0.0055,
+      "step": 63330
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.41375925925926e-05,
+      "loss": 0.0027,
+      "step": 63340
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.4136666666666674e-05,
+      "loss": 0.0049,
+      "step": 63350
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.413574074074074e-05,
+      "loss": 0.0036,
+      "step": 63360
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.413481481481482e-05,
+      "loss": 0.0024,
+      "step": 63370
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.413388888888889e-05,
+      "loss": 0.0035,
+      "step": 63380
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.413296296296297e-05,
+      "loss": 0.0031,
+      "step": 63390
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.4132037037037036e-05,
+      "loss": 0.005,
+      "step": 63400
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.413111111111111e-05,
+      "loss": 0.0041,
+      "step": 63410
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.4130185185185186e-05,
+      "loss": 0.0025,
+      "step": 63420
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.412925925925926e-05,
+      "loss": 0.0031,
+      "step": 63430
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.4128333333333336e-05,
+      "loss": 0.0079,
+      "step": 63440
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.412740740740741e-05,
+      "loss": 0.0036,
+      "step": 63450
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.4126481481481486e-05,
+      "loss": 0.0021,
+      "step": 63460
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.412555555555556e-05,
+      "loss": 0.004,
+      "step": 63470
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.412462962962963e-05,
+      "loss": 0.0085,
+      "step": 63480
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.4123703703703704e-05,
+      "loss": 0.0028,
+      "step": 63490
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.412277777777778e-05,
+      "loss": 0.0037,
+      "step": 63500
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.4121851851851854e-05,
+      "loss": 0.003,
+      "step": 63510
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.412092592592593e-05,
+      "loss": 0.0041,
+      "step": 63520
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.412e-05,
+      "loss": 0.0029,
+      "step": 63530
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.411907407407407e-05,
+      "loss": 0.0026,
+      "step": 63540
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.4118148148148155e-05,
+      "loss": 0.0027,
+      "step": 63550
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.411722222222223e-05,
+      "loss": 0.0048,
+      "step": 63560
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.41162962962963e-05,
+      "loss": 0.0039,
+      "step": 63570
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.411537037037037e-05,
+      "loss": 0.0039,
+      "step": 63580
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.411444444444445e-05,
+      "loss": 0.0038,
+      "step": 63590
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.411351851851852e-05,
+      "loss": 0.0024,
+      "step": 63600
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.411259259259259e-05,
+      "loss": 0.0036,
+      "step": 63610
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.4111666666666666e-05,
+      "loss": 0.0024,
+      "step": 63620
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.411074074074074e-05,
+      "loss": 0.0061,
+      "step": 63630
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.4109814814814817e-05,
+      "loss": 0.0026,
+      "step": 63640
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.410888888888889e-05,
+      "loss": 0.0041,
+      "step": 63650
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.410796296296297e-05,
+      "loss": 0.0055,
+      "step": 63660
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.410703703703704e-05,
+      "loss": 0.0025,
+      "step": 63670
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.410611111111112e-05,
+      "loss": 0.0038,
+      "step": 63680
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.4105185185185185e-05,
+      "loss": 0.0026,
+      "step": 63690
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.410425925925926e-05,
+      "loss": 0.0024,
+      "step": 63700
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.4103333333333335e-05,
+      "loss": 0.0049,
+      "step": 63710
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.410240740740741e-05,
+      "loss": 0.0057,
+      "step": 63720
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.4101481481481485e-05,
+      "loss": 0.005,
+      "step": 63730
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.4100555555555554e-05,
+      "loss": 0.0025,
+      "step": 63740
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.409962962962963e-05,
+      "loss": 0.0027,
+      "step": 63750
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.4098703703703704e-05,
+      "loss": 0.0017,
+      "step": 63760
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.4097777777777785e-05,
+      "loss": 0.0027,
+      "step": 63770
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.4096851851851854e-05,
+      "loss": 0.0043,
+      "step": 63780
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.409592592592593e-05,
+      "loss": 0.0038,
+      "step": 63790
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.4095000000000004e-05,
+      "loss": 0.0031,
+      "step": 63800
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.409407407407408e-05,
+      "loss": 0.0032,
+      "step": 63810
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.409314814814815e-05,
+      "loss": 0.0048,
+      "step": 63820
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.409222222222222e-05,
+      "loss": 0.0033,
+      "step": 63830
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.40912962962963e-05,
+      "loss": 0.0038,
+      "step": 63840
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.409037037037037e-05,
+      "loss": 0.0034,
+      "step": 63850
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.408944444444445e-05,
+      "loss": 0.0039,
+      "step": 63860
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.4088518518518516e-05,
+      "loss": 0.003,
+      "step": 63870
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.40875925925926e-05,
+      "loss": 0.0041,
+      "step": 63880
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.408666666666667e-05,
+      "loss": 0.0024,
+      "step": 63890
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.408574074074074e-05,
+      "loss": 0.0023,
+      "step": 63900
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.4084814814814816e-05,
+      "loss": 0.0027,
+      "step": 63910
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.408388888888889e-05,
+      "loss": 0.0031,
+      "step": 63920
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.4082962962962966e-05,
+      "loss": 0.0029,
+      "step": 63930
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.408203703703704e-05,
+      "loss": 0.0022,
+      "step": 63940
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.408111111111111e-05,
+      "loss": 0.0021,
+      "step": 63950
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.4080185185185184e-05,
+      "loss": 0.004,
+      "step": 63960
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.407925925925926e-05,
+      "loss": 0.0023,
+      "step": 63970
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.407833333333334e-05,
+      "loss": 0.0072,
+      "step": 63980
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.407740740740741e-05,
+      "loss": 0.0021,
+      "step": 63990
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.4076481481481485e-05,
+      "loss": 0.0013,
+      "step": 64000
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.407555555555556e-05,
+      "loss": 0.0038,
+      "step": 64010
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.4074629629629635e-05,
+      "loss": 0.0018,
+      "step": 64020
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.40737037037037e-05,
+      "loss": 0.003,
+      "step": 64030
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.407277777777778e-05,
+      "loss": 0.0026,
+      "step": 64040
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.407185185185185e-05,
+      "loss": 0.0034,
+      "step": 64050
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.407092592592593e-05,
+      "loss": 0.0024,
+      "step": 64060
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.407e-05,
+      "loss": 0.0039,
+      "step": 64070
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.406907407407407e-05,
+      "loss": 0.0041,
+      "step": 64080
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.406814814814815e-05,
+      "loss": 0.0022,
+      "step": 64090
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.406722222222223e-05,
+      "loss": 0.0031,
+      "step": 64100
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.40662962962963e-05,
+      "loss": 0.0034,
+      "step": 64110
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.406537037037037e-05,
+      "loss": 0.0022,
+      "step": 64120
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.406444444444445e-05,
+      "loss": 0.0044,
+      "step": 64130
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.406351851851852e-05,
+      "loss": 0.0047,
+      "step": 64140
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.40625925925926e-05,
+      "loss": 0.0046,
+      "step": 64150
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.4061666666666665e-05,
+      "loss": 0.0033,
+      "step": 64160
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.406074074074074e-05,
+      "loss": 0.0041,
+      "step": 64170
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.4059814814814815e-05,
+      "loss": 0.0036,
+      "step": 64180
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.405888888888889e-05,
+      "loss": 0.0045,
+      "step": 64190
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.4057962962962965e-05,
+      "loss": 0.0027,
+      "step": 64200
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.405703703703704e-05,
+      "loss": 0.0026,
+      "step": 64210
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.4056111111111116e-05,
+      "loss": 0.0028,
+      "step": 64220
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.405518518518519e-05,
+      "loss": 0.0072,
+      "step": 64230
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.4054351851851855e-05,
+      "loss": 0.0063,
+      "step": 64240
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.405342592592593e-05,
+      "loss": 0.0033,
+      "step": 64250
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.40525e-05,
+      "loss": 0.0056,
+      "step": 64260
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.4051574074074074e-05,
+      "loss": 0.0027,
+      "step": 64270
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.405064814814815e-05,
+      "loss": 0.0028,
+      "step": 64280
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.4049722222222224e-05,
+      "loss": 0.0046,
+      "step": 64290
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.40487962962963e-05,
+      "loss": 0.005,
+      "step": 64300
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.404787037037037e-05,
+      "loss": 0.0057,
+      "step": 64310
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.404694444444445e-05,
+      "loss": 0.004,
+      "step": 64320
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.4046018518518524e-05,
+      "loss": 0.0032,
+      "step": 64330
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.40450925925926e-05,
+      "loss": 0.0034,
+      "step": 64340
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.404416666666667e-05,
+      "loss": 0.0026,
+      "step": 64350
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.404324074074074e-05,
+      "loss": 0.0025,
+      "step": 64360
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.404231481481482e-05,
+      "loss": 0.0017,
+      "step": 64370
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.404138888888889e-05,
+      "loss": 0.0054,
+      "step": 64380
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.404046296296296e-05,
+      "loss": 0.0034,
+      "step": 64390
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.4039537037037036e-05,
+      "loss": 0.0028,
+      "step": 64400
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.403861111111111e-05,
+      "loss": 0.0039,
+      "step": 64410
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.4037685185185186e-05,
+      "loss": 0.0028,
+      "step": 64420
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.403675925925926e-05,
+      "loss": 0.004,
+      "step": 64430
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.4035833333333336e-05,
+      "loss": 0.0042,
+      "step": 64440
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.403490740740741e-05,
+      "loss": 0.0041,
+      "step": 64450
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.4033981481481486e-05,
+      "loss": 0.005,
+      "step": 64460
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.4033055555555555e-05,
+      "loss": 0.0041,
+      "step": 64470
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.403212962962963e-05,
+      "loss": 0.0026,
+      "step": 64480
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.4031203703703705e-05,
+      "loss": 0.0067,
+      "step": 64490
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.403027777777778e-05,
+      "loss": 0.0022,
+      "step": 64500
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.4029351851851855e-05,
+      "loss": 0.0037,
+      "step": 64510
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.402842592592592e-05,
+      "loss": 0.0049,
+      "step": 64520
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.4027500000000005e-05,
+      "loss": 0.0022,
+      "step": 64530
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.402657407407408e-05,
+      "loss": 0.0028,
+      "step": 64540
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.4025648148148155e-05,
+      "loss": 0.0033,
+      "step": 64550
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.402472222222222e-05,
+      "loss": 0.0058,
+      "step": 64560
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.40237962962963e-05,
+      "loss": 0.0042,
+      "step": 64570
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.4022870370370373e-05,
+      "loss": 0.0018,
+      "step": 64580
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.402194444444445e-05,
+      "loss": 0.0032,
+      "step": 64590
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.402101851851852e-05,
+      "loss": 0.0016,
+      "step": 64600
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.402009259259259e-05,
+      "loss": 0.0037,
+      "step": 64610
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.401916666666667e-05,
+      "loss": 0.0035,
+      "step": 64620
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.401824074074074e-05,
+      "loss": 0.0037,
+      "step": 64630
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.401731481481482e-05,
+      "loss": 0.0024,
+      "step": 64640
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.401638888888889e-05,
+      "loss": 0.0024,
+      "step": 64650
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.401546296296297e-05,
+      "loss": 0.0072,
+      "step": 64660
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.401453703703704e-05,
+      "loss": 0.002,
+      "step": 64670
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.401361111111111e-05,
+      "loss": 0.0024,
+      "step": 64680
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.4012685185185186e-05,
+      "loss": 0.0037,
+      "step": 64690
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.401175925925926e-05,
+      "loss": 0.0035,
+      "step": 64700
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.4010833333333336e-05,
+      "loss": 0.004,
+      "step": 64710
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.400990740740741e-05,
+      "loss": 0.0023,
+      "step": 64720
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.400898148148148e-05,
+      "loss": 0.0016,
+      "step": 64730
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.4008055555555554e-05,
+      "loss": 0.0042,
+      "step": 64740
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.4007129629629636e-05,
+      "loss": 0.0024,
+      "step": 64750
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.400620370370371e-05,
+      "loss": 0.0038,
+      "step": 64760
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.400527777777778e-05,
+      "loss": 0.0047,
+      "step": 64770
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.4004351851851854e-05,
+      "loss": 0.0038,
+      "step": 64780
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.400342592592593e-05,
+      "loss": 0.0033,
+      "step": 64790
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.4002500000000004e-05,
+      "loss": 0.0025,
+      "step": 64800
+    },
+    {
+      "epoch": 12.0,
+      "eval_loss": 0.014796755276620388,
+      "eval_runtime": 153.5613,
+      "eval_samples_per_second": 25.983,
+      "eval_steps_per_second": 3.25,
+      "step": 64800
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.400157407407407e-05,
+      "loss": 0.0023,
+      "step": 64810
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.400064814814815e-05,
+      "loss": 0.0027,
+      "step": 64820
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.399972222222222e-05,
+      "loss": 0.0044,
+      "step": 64830
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.39987962962963e-05,
+      "loss": 0.0033,
+      "step": 64840
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.399787037037037e-05,
+      "loss": 0.0041,
+      "step": 64850
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.399694444444445e-05,
+      "loss": 0.0022,
+      "step": 64860
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.399601851851852e-05,
+      "loss": 0.0022,
+      "step": 64870
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.39950925925926e-05,
+      "loss": 0.0019,
+      "step": 64880
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.3994166666666666e-05,
+      "loss": 0.0027,
+      "step": 64890
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.399324074074074e-05,
+      "loss": 0.0023,
+      "step": 64900
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.3992314814814816e-05,
+      "loss": 0.0036,
+      "step": 64910
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.399138888888889e-05,
+      "loss": 0.0027,
+      "step": 64920
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.3990462962962967e-05,
+      "loss": 0.0033,
+      "step": 64930
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.3989537037037035e-05,
+      "loss": 0.0023,
+      "step": 64940
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.398861111111111e-05,
+      "loss": 0.0026,
+      "step": 64950
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.3987685185185185e-05,
+      "loss": 0.0026,
+      "step": 64960
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.398675925925927e-05,
+      "loss": 0.0018,
+      "step": 64970
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.3985833333333335e-05,
+      "loss": 0.0033,
+      "step": 64980
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.398490740740741e-05,
+      "loss": 0.0037,
+      "step": 64990
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.3983981481481485e-05,
+      "loss": 0.0025,
+      "step": 65000
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.398305555555556e-05,
+      "loss": 0.0025,
+      "step": 65010
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.398212962962963e-05,
+      "loss": 0.0038,
+      "step": 65020
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.3981203703703704e-05,
+      "loss": 0.002,
+      "step": 65030
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.398027777777778e-05,
+      "loss": 0.003,
+      "step": 65040
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.3979351851851854e-05,
+      "loss": 0.0019,
+      "step": 65050
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.397842592592593e-05,
+      "loss": 0.0029,
+      "step": 65060
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.3977500000000004e-05,
+      "loss": 0.0019,
+      "step": 65070
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.397657407407408e-05,
+      "loss": 0.0032,
+      "step": 65080
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.3975648148148154e-05,
+      "loss": 0.005,
+      "step": 65090
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.397472222222222e-05,
+      "loss": 0.0034,
+      "step": 65100
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.39737962962963e-05,
+      "loss": 0.0031,
+      "step": 65110
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.397287037037037e-05,
+      "loss": 0.0043,
+      "step": 65120
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.397194444444445e-05,
+      "loss": 0.0012,
+      "step": 65130
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.397101851851852e-05,
+      "loss": 0.0018,
+      "step": 65140
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.397009259259259e-05,
+      "loss": 0.0042,
+      "step": 65150
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.3969166666666666e-05,
+      "loss": 0.0018,
+      "step": 65160
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.396824074074074e-05,
+      "loss": 0.0034,
+      "step": 65170
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.396731481481482e-05,
+      "loss": 0.0043,
+      "step": 65180
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.396638888888889e-05,
+      "loss": 0.0024,
+      "step": 65190
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.3965462962962966e-05,
+      "loss": 0.0045,
+      "step": 65200
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.396453703703704e-05,
+      "loss": 0.0021,
+      "step": 65210
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.3963611111111116e-05,
+      "loss": 0.0035,
+      "step": 65220
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.3962685185185184e-05,
+      "loss": 0.0028,
+      "step": 65230
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.396175925925926e-05,
+      "loss": 0.0024,
+      "step": 65240
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.3960833333333334e-05,
+      "loss": 0.0025,
+      "step": 65250
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.395990740740741e-05,
+      "loss": 0.0021,
+      "step": 65260
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.3958981481481485e-05,
+      "loss": 0.0013,
+      "step": 65270
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.395805555555555e-05,
+      "loss": 0.0046,
+      "step": 65280
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.3957129629629635e-05,
+      "loss": 0.0021,
+      "step": 65290
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.395620370370371e-05,
+      "loss": 0.0034,
+      "step": 65300
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.395527777777778e-05,
+      "loss": 0.0037,
+      "step": 65310
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.395435185185185e-05,
+      "loss": 0.0036,
+      "step": 65320
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.3953518518518525e-05,
+      "loss": 0.0064,
+      "step": 65330
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.395259259259259e-05,
+      "loss": 0.0034,
+      "step": 65340
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.395166666666667e-05,
+      "loss": 0.0043,
+      "step": 65350
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.395074074074074e-05,
+      "loss": 0.002,
+      "step": 65360
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.394981481481482e-05,
+      "loss": 0.0082,
+      "step": 65370
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.3948888888888886e-05,
+      "loss": 0.0037,
+      "step": 65380
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.394796296296296e-05,
+      "loss": 0.0025,
+      "step": 65390
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.3947037037037036e-05,
+      "loss": 0.0025,
+      "step": 65400
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.394611111111112e-05,
+      "loss": 0.0033,
+      "step": 65410
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.3945185185185187e-05,
+      "loss": 0.0014,
+      "step": 65420
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.394425925925926e-05,
+      "loss": 0.0015,
+      "step": 65430
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.394333333333334e-05,
+      "loss": 0.0037,
+      "step": 65440
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.394240740740741e-05,
+      "loss": 0.0026,
+      "step": 65450
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.394148148148149e-05,
+      "loss": 0.0033,
+      "step": 65460
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.3940555555555555e-05,
+      "loss": 0.0033,
+      "step": 65470
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.393962962962963e-05,
+      "loss": 0.0023,
+      "step": 65480
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.3938703703703705e-05,
+      "loss": 0.0038,
+      "step": 65490
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.393777777777778e-05,
+      "loss": 0.0024,
+      "step": 65500
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.393685185185185e-05,
+      "loss": 0.0061,
+      "step": 65510
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.393592592592593e-05,
+      "loss": 0.0023,
+      "step": 65520
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.3935000000000005e-05,
+      "loss": 0.0034,
+      "step": 65530
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.393407407407408e-05,
+      "loss": 0.0072,
+      "step": 65540
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.393314814814815e-05,
+      "loss": 0.0027,
+      "step": 65550
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.3932222222222224e-05,
+      "loss": 0.002,
+      "step": 65560
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.39312962962963e-05,
+      "loss": 0.0049,
+      "step": 65570
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.3930370370370374e-05,
+      "loss": 0.0032,
+      "step": 65580
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.392944444444444e-05,
+      "loss": 0.005,
+      "step": 65590
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.392851851851852e-05,
+      "loss": 0.0038,
+      "step": 65600
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.392759259259259e-05,
+      "loss": 0.0018,
+      "step": 65610
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.3926666666666674e-05,
+      "loss": 0.0018,
+      "step": 65620
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.392574074074074e-05,
+      "loss": 0.0043,
+      "step": 65630
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.392481481481482e-05,
+      "loss": 0.0025,
+      "step": 65640
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.392388888888889e-05,
+      "loss": 0.0027,
+      "step": 65650
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.392296296296297e-05,
+      "loss": 0.0025,
+      "step": 65660
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.392203703703704e-05,
+      "loss": 0.0035,
+      "step": 65670
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.392111111111111e-05,
+      "loss": 0.0024,
+      "step": 65680
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.3920185185185186e-05,
+      "loss": 0.0043,
+      "step": 65690
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.391925925925926e-05,
+      "loss": 0.0018,
+      "step": 65700
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.3918333333333336e-05,
+      "loss": 0.0026,
+      "step": 65710
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.3917407407407404e-05,
+      "loss": 0.0046,
+      "step": 65720
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.3916481481481486e-05,
+      "loss": 0.0046,
+      "step": 65730
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.391555555555556e-05,
+      "loss": 0.0027,
+      "step": 65740
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.3914629629629636e-05,
+      "loss": 0.0019,
+      "step": 65750
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.3913703703703705e-05,
+      "loss": 0.0044,
+      "step": 65760
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.391277777777778e-05,
+      "loss": 0.0037,
+      "step": 65770
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.3911851851851855e-05,
+      "loss": 0.0035,
+      "step": 65780
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.391092592592593e-05,
+      "loss": 0.0042,
+      "step": 65790
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.391e-05,
+      "loss": 0.0026,
+      "step": 65800
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.390907407407407e-05,
+      "loss": 0.003,
+      "step": 65810
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.390814814814815e-05,
+      "loss": 0.0025,
+      "step": 65820
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.390722222222222e-05,
+      "loss": 0.0036,
+      "step": 65830
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.39062962962963e-05,
+      "loss": 0.0042,
+      "step": 65840
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.390537037037037e-05,
+      "loss": 0.0023,
+      "step": 65850
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.390444444444445e-05,
+      "loss": 0.0027,
+      "step": 65860
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.3903518518518523e-05,
+      "loss": 0.0049,
+      "step": 65870
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.39025925925926e-05,
+      "loss": 0.0032,
+      "step": 65880
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.390166666666667e-05,
+      "loss": 0.0037,
+      "step": 65890
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.390074074074074e-05,
+      "loss": 0.0022,
+      "step": 65900
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.389981481481482e-05,
+      "loss": 0.0042,
+      "step": 65910
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.389888888888889e-05,
+      "loss": 0.0013,
+      "step": 65920
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.389796296296296e-05,
+      "loss": 0.0049,
+      "step": 65930
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.3897037037037035e-05,
+      "loss": 0.0024,
+      "step": 65940
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.389611111111112e-05,
+      "loss": 0.0021,
+      "step": 65950
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.389518518518519e-05,
+      "loss": 0.002,
+      "step": 65960
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.389425925925926e-05,
+      "loss": 0.0036,
+      "step": 65970
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.3893333333333335e-05,
+      "loss": 0.0028,
+      "step": 65980
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.389240740740741e-05,
+      "loss": 0.0042,
+      "step": 65990
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.3891481481481486e-05,
+      "loss": 0.0018,
+      "step": 66000
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.3890555555555554e-05,
+      "loss": 0.0032,
+      "step": 66010
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.388962962962963e-05,
+      "loss": 0.003,
+      "step": 66020
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.3888703703703704e-05,
+      "loss": 0.0015,
+      "step": 66030
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.388777777777778e-05,
+      "loss": 0.0025,
+      "step": 66040
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.3886851851851854e-05,
+      "loss": 0.0035,
+      "step": 66050
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.388592592592593e-05,
+      "loss": 0.0029,
+      "step": 66060
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.3885000000000004e-05,
+      "loss": 0.004,
+      "step": 66070
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.388407407407408e-05,
+      "loss": 0.0015,
+      "step": 66080
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.3883148148148154e-05,
+      "loss": 0.0028,
+      "step": 66090
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.388222222222222e-05,
+      "loss": 0.0031,
+      "step": 66100
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.38812962962963e-05,
+      "loss": 0.0047,
+      "step": 66110
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.388037037037037e-05,
+      "loss": 0.0023,
+      "step": 66120
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.387944444444445e-05,
+      "loss": 0.002,
+      "step": 66130
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.3878518518518516e-05,
+      "loss": 0.0037,
+      "step": 66140
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.387759259259259e-05,
+      "loss": 0.0042,
+      "step": 66150
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.387666666666667e-05,
+      "loss": 0.0024,
+      "step": 66160
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.387574074074075e-05,
+      "loss": 0.0055,
+      "step": 66170
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.3874814814814816e-05,
+      "loss": 0.0034,
+      "step": 66180
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.387388888888889e-05,
+      "loss": 0.0062,
+      "step": 66190
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.3872962962962966e-05,
+      "loss": 0.0057,
+      "step": 66200
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.387203703703704e-05,
+      "loss": 0.0026,
+      "step": 66210
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.387111111111111e-05,
+      "loss": 0.0034,
+      "step": 66220
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.3870185185185185e-05,
+      "loss": 0.0075,
+      "step": 66230
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.386925925925926e-05,
+      "loss": 0.0031,
+      "step": 66240
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.3868333333333335e-05,
+      "loss": 0.0034,
+      "step": 66250
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.386740740740741e-05,
+      "loss": 0.0028,
+      "step": 66260
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.3866481481481485e-05,
+      "loss": 0.0023,
+      "step": 66270
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.386555555555556e-05,
+      "loss": 0.002,
+      "step": 66280
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.3864629629629635e-05,
+      "loss": 0.0052,
+      "step": 66290
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.386370370370371e-05,
+      "loss": 0.0052,
+      "step": 66300
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.386277777777778e-05,
+      "loss": 0.0043,
+      "step": 66310
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.3861851851851853e-05,
+      "loss": 0.0045,
+      "step": 66320
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.386092592592593e-05,
+      "loss": 0.0013,
+      "step": 66330
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.3860000000000004e-05,
+      "loss": 0.002,
+      "step": 66340
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.385907407407407e-05,
+      "loss": 0.0025,
+      "step": 66350
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.385814814814815e-05,
+      "loss": 0.0029,
+      "step": 66360
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.385722222222222e-05,
+      "loss": 0.0027,
+      "step": 66370
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.3856296296296304e-05,
+      "loss": 0.0042,
+      "step": 66380
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.385537037037037e-05,
+      "loss": 0.0017,
+      "step": 66390
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.385444444444445e-05,
+      "loss": 0.0059,
+      "step": 66400
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.385351851851852e-05,
+      "loss": 0.0059,
+      "step": 66410
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.38525925925926e-05,
+      "loss": 0.0048,
+      "step": 66420
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.3851666666666666e-05,
+      "loss": 0.0046,
+      "step": 66430
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.385074074074074e-05,
+      "loss": 0.002,
+      "step": 66440
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.3849814814814816e-05,
+      "loss": 0.0028,
+      "step": 66450
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.384888888888889e-05,
+      "loss": 0.0042,
+      "step": 66460
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.3847962962962966e-05,
+      "loss": 0.0033,
+      "step": 66470
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.3847037037037034e-05,
+      "loss": 0.0035,
+      "step": 66480
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.3846111111111116e-05,
+      "loss": 0.0014,
+      "step": 66490
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.384518518518519e-05,
+      "loss": 0.0031,
+      "step": 66500
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.3844259259259266e-05,
+      "loss": 0.0033,
+      "step": 66510
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.3843333333333334e-05,
+      "loss": 0.0043,
+      "step": 66520
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.384240740740741e-05,
+      "loss": 0.0023,
+      "step": 66530
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.3841481481481484e-05,
+      "loss": 0.0045,
+      "step": 66540
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.384055555555556e-05,
+      "loss": 0.0029,
+      "step": 66550
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.383962962962963e-05,
+      "loss": 0.002,
+      "step": 66560
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.38387037037037e-05,
+      "loss": 0.0035,
+      "step": 66570
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.383777777777778e-05,
+      "loss": 0.0026,
+      "step": 66580
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.383685185185186e-05,
+      "loss": 0.0029,
+      "step": 66590
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.383592592592593e-05,
+      "loss": 0.0028,
+      "step": 66600
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.3835e-05,
+      "loss": 0.0016,
+      "step": 66610
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.383407407407408e-05,
+      "loss": 0.0034,
+      "step": 66620
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.383314814814815e-05,
+      "loss": 0.0041,
+      "step": 66630
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.383222222222222e-05,
+      "loss": 0.0041,
+      "step": 66640
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.3831296296296296e-05,
+      "loss": 0.0027,
+      "step": 66650
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.383037037037037e-05,
+      "loss": 0.0029,
+      "step": 66660
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.3829444444444447e-05,
+      "loss": 0.0032,
+      "step": 66670
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.382851851851852e-05,
+      "loss": 0.0021,
+      "step": 66680
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.382759259259259e-05,
+      "loss": 0.003,
+      "step": 66690
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.382666666666667e-05,
+      "loss": 0.0028,
+      "step": 66700
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.382574074074075e-05,
+      "loss": 0.0045,
+      "step": 66710
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.382481481481482e-05,
+      "loss": 0.0038,
+      "step": 66720
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.382388888888889e-05,
+      "loss": 0.0041,
+      "step": 66730
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.3822962962962965e-05,
+      "loss": 0.0037,
+      "step": 66740
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.382203703703704e-05,
+      "loss": 0.0031,
+      "step": 66750
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.3821111111111115e-05,
+      "loss": 0.0032,
+      "step": 66760
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.3820185185185184e-05,
+      "loss": 0.0024,
+      "step": 66770
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.381925925925926e-05,
+      "loss": 0.0032,
+      "step": 66780
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.3818333333333334e-05,
+      "loss": 0.0037,
+      "step": 66790
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.381740740740741e-05,
+      "loss": 0.0023,
+      "step": 66800
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.3816481481481484e-05,
+      "loss": 0.0026,
+      "step": 66810
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.381555555555556e-05,
+      "loss": 0.0036,
+      "step": 66820
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.3814629629629634e-05,
+      "loss": 0.0044,
+      "step": 66830
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.381370370370371e-05,
+      "loss": 0.0014,
+      "step": 66840
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.381277777777778e-05,
+      "loss": 0.0042,
+      "step": 66850
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.381185185185185e-05,
+      "loss": 0.0028,
+      "step": 66860
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.381092592592593e-05,
+      "loss": 0.0049,
+      "step": 66870
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.381e-05,
+      "loss": 0.0014,
+      "step": 66880
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.380907407407408e-05,
+      "loss": 0.0019,
+      "step": 66890
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.3808148148148146e-05,
+      "loss": 0.003,
+      "step": 66900
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.380722222222222e-05,
+      "loss": 0.0057,
+      "step": 66910
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.38062962962963e-05,
+      "loss": 0.0029,
+      "step": 66920
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.380537037037038e-05,
+      "loss": 0.0031,
+      "step": 66930
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.3804444444444446e-05,
+      "loss": 0.0037,
+      "step": 66940
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.380351851851852e-05,
+      "loss": 0.0016,
+      "step": 66950
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.3802592592592596e-05,
+      "loss": 0.0044,
+      "step": 66960
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.380166666666667e-05,
+      "loss": 0.0017,
+      "step": 66970
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.380074074074074e-05,
+      "loss": 0.0033,
+      "step": 66980
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.3799814814814814e-05,
+      "loss": 0.0032,
+      "step": 66990
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.379888888888889e-05,
+      "loss": 0.0022,
+      "step": 67000
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.3797962962962965e-05,
+      "loss": 0.0035,
+      "step": 67010
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.379703703703704e-05,
+      "loss": 0.0037,
+      "step": 67020
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.3796111111111115e-05,
+      "loss": 0.0024,
+      "step": 67030
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.379518518518519e-05,
+      "loss": 0.0013,
+      "step": 67040
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.3794259259259265e-05,
+      "loss": 0.0036,
+      "step": 67050
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.379333333333333e-05,
+      "loss": 0.0031,
+      "step": 67060
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.379240740740741e-05,
+      "loss": 0.0023,
+      "step": 67070
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.379148148148148e-05,
+      "loss": 0.0027,
+      "step": 67080
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.379055555555556e-05,
+      "loss": 0.0023,
+      "step": 67090
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.378962962962963e-05,
+      "loss": 0.0028,
+      "step": 67100
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.37887037037037e-05,
+      "loss": 0.0027,
+      "step": 67110
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.3787777777777777e-05,
+      "loss": 0.0027,
+      "step": 67120
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.378685185185186e-05,
+      "loss": 0.0026,
+      "step": 67130
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.3785925925925933e-05,
+      "loss": 0.0043,
+      "step": 67140
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.3785e-05,
+      "loss": 0.003,
+      "step": 67150
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.378407407407408e-05,
+      "loss": 0.0039,
+      "step": 67160
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.378314814814815e-05,
+      "loss": 0.0019,
+      "step": 67170
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.378222222222223e-05,
+      "loss": 0.0036,
+      "step": 67180
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.3781296296296295e-05,
+      "loss": 0.0042,
+      "step": 67190
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.378037037037037e-05,
+      "loss": 0.0027,
+      "step": 67200
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.3779444444444445e-05,
+      "loss": 0.0026,
+      "step": 67210
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.377851851851852e-05,
+      "loss": 0.0042,
+      "step": 67220
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.3777592592592595e-05,
+      "loss": 0.0035,
+      "step": 67230
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.377666666666667e-05,
+      "loss": 0.0037,
+      "step": 67240
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.3775740740740746e-05,
+      "loss": 0.002,
+      "step": 67250
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.377481481481482e-05,
+      "loss": 0.0027,
+      "step": 67260
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.377388888888889e-05,
+      "loss": 0.0027,
+      "step": 67270
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.3772962962962964e-05,
+      "loss": 0.003,
+      "step": 67280
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.377203703703704e-05,
+      "loss": 0.0039,
+      "step": 67290
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.3771111111111114e-05,
+      "loss": 0.0031,
+      "step": 67300
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.377018518518519e-05,
+      "loss": 0.0017,
+      "step": 67310
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.376925925925926e-05,
+      "loss": 0.0045,
+      "step": 67320
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.376833333333333e-05,
+      "loss": 0.0032,
+      "step": 67330
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.376740740740741e-05,
+      "loss": 0.0024,
+      "step": 67340
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.376648148148149e-05,
+      "loss": 0.0036,
+      "step": 67350
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.376555555555556e-05,
+      "loss": 0.0027,
+      "step": 67360
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.376462962962963e-05,
+      "loss": 0.0023,
+      "step": 67370
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.376370370370371e-05,
+      "loss": 0.004,
+      "step": 67380
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.376277777777778e-05,
+      "loss": 0.002,
+      "step": 67390
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.376185185185185e-05,
+      "loss": 0.0038,
+      "step": 67400
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.3760925925925926e-05,
+      "loss": 0.003,
+      "step": 67410
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.376e-05,
+      "loss": 0.0036,
+      "step": 67420
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.3759074074074076e-05,
+      "loss": 0.0024,
+      "step": 67430
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.375814814814815e-05,
+      "loss": 0.0039,
+      "step": 67440
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.375722222222222e-05,
+      "loss": 0.0035,
+      "step": 67450
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.37562962962963e-05,
+      "loss": 0.0057,
+      "step": 67460
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.3755370370370376e-05,
+      "loss": 0.0047,
+      "step": 67470
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.3754444444444445e-05,
+      "loss": 0.0026,
+      "step": 67480
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.375351851851852e-05,
+      "loss": 0.0025,
+      "step": 67490
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.3752592592592595e-05,
+      "loss": 0.0035,
+      "step": 67500
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.375166666666667e-05,
+      "loss": 0.0029,
+      "step": 67510
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.3750740740740745e-05,
+      "loss": 0.0051,
+      "step": 67520
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.374981481481481e-05,
+      "loss": 0.0028,
+      "step": 67530
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.374888888888889e-05,
+      "loss": 0.0017,
+      "step": 67540
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.374796296296296e-05,
+      "loss": 0.0021,
+      "step": 67550
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.374703703703704e-05,
+      "loss": 0.0031,
+      "step": 67560
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.3746111111111113e-05,
+      "loss": 0.0036,
+      "step": 67570
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.374518518518519e-05,
+      "loss": 0.0047,
+      "step": 67580
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.3744259259259264e-05,
+      "loss": 0.0024,
+      "step": 67590
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.374333333333334e-05,
+      "loss": 0.0032,
+      "step": 67600
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.374240740740741e-05,
+      "loss": 0.0024,
+      "step": 67610
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.374148148148148e-05,
+      "loss": 0.0037,
+      "step": 67620
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.374055555555556e-05,
+      "loss": 0.0045,
+      "step": 67630
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.373962962962963e-05,
+      "loss": 0.0034,
+      "step": 67640
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.373870370370371e-05,
+      "loss": 0.002,
+      "step": 67650
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.3737777777777775e-05,
+      "loss": 0.0033,
+      "step": 67660
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.373685185185186e-05,
+      "loss": 0.0026,
+      "step": 67670
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.373592592592593e-05,
+      "loss": 0.0029,
+      "step": 67680
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.3735e-05,
+      "loss": 0.002,
+      "step": 67690
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.3734074074074076e-05,
+      "loss": 0.0028,
+      "step": 67700
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.373314814814815e-05,
+      "loss": 0.003,
+      "step": 67710
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.3732222222222226e-05,
+      "loss": 0.0047,
+      "step": 67720
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.37312962962963e-05,
+      "loss": 0.0067,
+      "step": 67730
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.373037037037037e-05,
+      "loss": 0.0015,
+      "step": 67740
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.3729444444444444e-05,
+      "loss": 0.002,
+      "step": 67750
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.372851851851852e-05,
+      "loss": 0.0061,
+      "step": 67760
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.3727592592592594e-05,
+      "loss": 0.0025,
+      "step": 67770
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.372666666666667e-05,
+      "loss": 0.0023,
+      "step": 67780
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.3725740740740744e-05,
+      "loss": 0.0046,
+      "step": 67790
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.372481481481482e-05,
+      "loss": 0.0073,
+      "step": 67800
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.3723888888888894e-05,
+      "loss": 0.0085,
+      "step": 67810
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.372296296296296e-05,
+      "loss": 0.0023,
+      "step": 67820
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.372203703703704e-05,
+      "loss": 0.0027,
+      "step": 67830
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.372111111111111e-05,
+      "loss": 0.0019,
+      "step": 67840
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.372018518518519e-05,
+      "loss": 0.0032,
+      "step": 67850
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.371925925925926e-05,
+      "loss": 0.0033,
+      "step": 67860
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.371833333333333e-05,
+      "loss": 0.0051,
+      "step": 67870
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.3717407407407406e-05,
+      "loss": 0.0053,
+      "step": 67880
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.371648148148149e-05,
+      "loss": 0.0051,
+      "step": 67890
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.3715555555555556e-05,
+      "loss": 0.0056,
+      "step": 67900
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.371462962962963e-05,
+      "loss": 0.0031,
+      "step": 67910
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.3713703703703706e-05,
+      "loss": 0.0027,
+      "step": 67920
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.371277777777778e-05,
+      "loss": 0.0033,
+      "step": 67930
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.3711851851851857e-05,
+      "loss": 0.004,
+      "step": 67940
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.3710925925925925e-05,
+      "loss": 0.0061,
+      "step": 67950
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.371e-05,
+      "loss": 0.002,
+      "step": 67960
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.3709074074074075e-05,
+      "loss": 0.0039,
+      "step": 67970
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.370814814814815e-05,
+      "loss": 0.0042,
+      "step": 67980
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.370722222222222e-05,
+      "loss": 0.0046,
+      "step": 67990
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.37062962962963e-05,
+      "loss": 0.0054,
+      "step": 68000
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.3705370370370375e-05,
+      "loss": 0.0016,
+      "step": 68010
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.370444444444445e-05,
+      "loss": 0.0031,
+      "step": 68020
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.370351851851852e-05,
+      "loss": 0.0048,
+      "step": 68030
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.3702592592592594e-05,
+      "loss": 0.0033,
+      "step": 68040
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.370166666666667e-05,
+      "loss": 0.002,
+      "step": 68050
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.3700740740740744e-05,
+      "loss": 0.0035,
+      "step": 68060
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.369981481481482e-05,
+      "loss": 0.003,
+      "step": 68070
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.369888888888889e-05,
+      "loss": 0.0049,
+      "step": 68080
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.369796296296296e-05,
+      "loss": 0.0041,
+      "step": 68090
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.3697037037037044e-05,
+      "loss": 0.0059,
+      "step": 68100
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.369611111111111e-05,
+      "loss": 0.0015,
+      "step": 68110
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.369518518518519e-05,
+      "loss": 0.0033,
+      "step": 68120
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.369425925925926e-05,
+      "loss": 0.0023,
+      "step": 68130
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.369333333333334e-05,
+      "loss": 0.0029,
+      "step": 68140
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.369240740740741e-05,
+      "loss": 0.0044,
+      "step": 68150
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.369148148148148e-05,
+      "loss": 0.0035,
+      "step": 68160
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.3690555555555556e-05,
+      "loss": 0.006,
+      "step": 68170
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.368962962962963e-05,
+      "loss": 0.0033,
+      "step": 68180
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.3688703703703706e-05,
+      "loss": 0.0031,
+      "step": 68190
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.3687777777777774e-05,
+      "loss": 0.0056,
+      "step": 68200
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.3686851851851856e-05,
+      "loss": 0.0042,
+      "step": 68210
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.368592592592593e-05,
+      "loss": 0.002,
+      "step": 68220
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.3685000000000006e-05,
+      "loss": 0.0047,
+      "step": 68230
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.3684074074074074e-05,
+      "loss": 0.0041,
+      "step": 68240
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.368314814814815e-05,
+      "loss": 0.0047,
+      "step": 68250
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.3682222222222224e-05,
+      "loss": 0.0035,
+      "step": 68260
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.36812962962963e-05,
+      "loss": 0.0065,
+      "step": 68270
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.3680370370370375e-05,
+      "loss": 0.0035,
+      "step": 68280
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.367944444444444e-05,
+      "loss": 0.0026,
+      "step": 68290
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.367851851851852e-05,
+      "loss": 0.0052,
+      "step": 68300
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.367759259259259e-05,
+      "loss": 0.0022,
+      "step": 68310
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.367666666666667e-05,
+      "loss": 0.0042,
+      "step": 68320
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.367574074074074e-05,
+      "loss": 0.0036,
+      "step": 68330
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.367481481481482e-05,
+      "loss": 0.004,
+      "step": 68340
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.367388888888889e-05,
+      "loss": 0.0033,
+      "step": 68350
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.367296296296297e-05,
+      "loss": 0.004,
+      "step": 68360
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.3672037037037037e-05,
+      "loss": 0.0023,
+      "step": 68370
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.367111111111111e-05,
+      "loss": 0.0026,
+      "step": 68380
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.367018518518519e-05,
+      "loss": 0.0056,
+      "step": 68390
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.366925925925926e-05,
+      "loss": 0.0024,
+      "step": 68400
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.366833333333333e-05,
+      "loss": 0.0028,
+      "step": 68410
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.3667407407407405e-05,
+      "loss": 0.0045,
+      "step": 68420
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.366648148148149e-05,
+      "loss": 0.0025,
+      "step": 68430
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.366555555555556e-05,
+      "loss": 0.0033,
+      "step": 68440
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.366462962962963e-05,
+      "loss": 0.0037,
+      "step": 68450
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.3663703703703705e-05,
+      "loss": 0.0037,
+      "step": 68460
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.366277777777778e-05,
+      "loss": 0.0038,
+      "step": 68470
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.3661851851851855e-05,
+      "loss": 0.0028,
+      "step": 68480
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.366092592592593e-05,
+      "loss": 0.0047,
+      "step": 68490
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.366e-05,
+      "loss": 0.0032,
+      "step": 68500
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.3659074074074074e-05,
+      "loss": 0.0025,
+      "step": 68510
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.365814814814815e-05,
+      "loss": 0.0028,
+      "step": 68520
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.3657222222222224e-05,
+      "loss": 0.0026,
+      "step": 68530
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.36562962962963e-05,
+      "loss": 0.0037,
+      "step": 68540
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.3655370370370374e-05,
+      "loss": 0.0041,
+      "step": 68550
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.365444444444445e-05,
+      "loss": 0.0032,
+      "step": 68560
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.3653518518518524e-05,
+      "loss": 0.0025,
+      "step": 68570
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.365259259259259e-05,
+      "loss": 0.003,
+      "step": 68580
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.365166666666667e-05,
+      "loss": 0.0064,
+      "step": 68590
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.365074074074074e-05,
+      "loss": 0.0028,
+      "step": 68600
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.364981481481482e-05,
+      "loss": 0.0032,
+      "step": 68610
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.3648888888888886e-05,
+      "loss": 0.0025,
+      "step": 68620
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.364796296296296e-05,
+      "loss": 0.0022,
+      "step": 68630
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.364703703703704e-05,
+      "loss": 0.002,
+      "step": 68640
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.364611111111112e-05,
+      "loss": 0.0032,
+      "step": 68650
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.3645185185185186e-05,
+      "loss": 0.0043,
+      "step": 68660
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.364425925925926e-05,
+      "loss": 0.0027,
+      "step": 68670
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.3643333333333336e-05,
+      "loss": 0.0043,
+      "step": 68680
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.364240740740741e-05,
+      "loss": 0.0021,
+      "step": 68690
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.3641481481481486e-05,
+      "loss": 0.0023,
+      "step": 68700
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.3640555555555555e-05,
+      "loss": 0.003,
+      "step": 68710
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.363962962962963e-05,
+      "loss": 0.0023,
+      "step": 68720
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.3638703703703705e-05,
+      "loss": 0.0032,
+      "step": 68730
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.363777777777778e-05,
+      "loss": 0.0026,
+      "step": 68740
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.3636851851851855e-05,
+      "loss": 0.0032,
+      "step": 68750
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.363592592592593e-05,
+      "loss": 0.0025,
+      "step": 68760
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.3635000000000005e-05,
+      "loss": 0.0032,
+      "step": 68770
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.363407407407408e-05,
+      "loss": 0.002,
+      "step": 68780
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.363314814814815e-05,
+      "loss": 0.0017,
+      "step": 68790
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.363222222222222e-05,
+      "loss": 0.002,
+      "step": 68800
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.36312962962963e-05,
+      "loss": 0.0022,
+      "step": 68810
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.3630370370370373e-05,
+      "loss": 0.0025,
+      "step": 68820
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.362944444444444e-05,
+      "loss": 0.0029,
+      "step": 68830
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.362851851851852e-05,
+      "loss": 0.0015,
+      "step": 68840
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.362759259259259e-05,
+      "loss": 0.0018,
+      "step": 68850
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.3626666666666674e-05,
+      "loss": 0.0016,
+      "step": 68860
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.362574074074074e-05,
+      "loss": 0.0035,
+      "step": 68870
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.362481481481482e-05,
+      "loss": 0.0018,
+      "step": 68880
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.362388888888889e-05,
+      "loss": 0.0022,
+      "step": 68890
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.362296296296297e-05,
+      "loss": 0.0033,
+      "step": 68900
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.362203703703704e-05,
+      "loss": 0.0034,
+      "step": 68910
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.362111111111111e-05,
+      "loss": 0.0021,
+      "step": 68920
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.3620185185185185e-05,
+      "loss": 0.0023,
+      "step": 68930
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.361925925925926e-05,
+      "loss": 0.0037,
+      "step": 68940
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.3618333333333336e-05,
+      "loss": 0.0041,
+      "step": 68950
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.3617407407407404e-05,
+      "loss": 0.0032,
+      "step": 68960
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.3616481481481486e-05,
+      "loss": 0.0069,
+      "step": 68970
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.361555555555556e-05,
+      "loss": 0.0034,
+      "step": 68980
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.3614629629629636e-05,
+      "loss": 0.0032,
+      "step": 68990
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.3613703703703704e-05,
+      "loss": 0.0012,
+      "step": 69000
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.361277777777778e-05,
+      "loss": 0.0028,
+      "step": 69010
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.3611851851851854e-05,
+      "loss": 0.0031,
+      "step": 69020
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.361092592592593e-05,
+      "loss": 0.0022,
+      "step": 69030
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.361e-05,
+      "loss": 0.0033,
+      "step": 69040
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.360907407407407e-05,
+      "loss": 0.0024,
+      "step": 69050
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.360814814814815e-05,
+      "loss": 0.0046,
+      "step": 69060
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.360722222222223e-05,
+      "loss": 0.0031,
+      "step": 69070
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.36062962962963e-05,
+      "loss": 0.0019,
+      "step": 69080
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.360537037037037e-05,
+      "loss": 0.0039,
+      "step": 69090
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.360444444444445e-05,
+      "loss": 0.0055,
+      "step": 69100
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.360351851851852e-05,
+      "loss": 0.0028,
+      "step": 69110
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.36025925925926e-05,
+      "loss": 0.0028,
+      "step": 69120
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.3601666666666666e-05,
+      "loss": 0.005,
+      "step": 69130
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.360074074074074e-05,
+      "loss": 0.0022,
+      "step": 69140
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.3599814814814816e-05,
+      "loss": 0.0019,
+      "step": 69150
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.359888888888889e-05,
+      "loss": 0.0034,
+      "step": 69160
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.359796296296296e-05,
+      "loss": 0.0048,
+      "step": 69170
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.359703703703704e-05,
+      "loss": 0.0022,
+      "step": 69180
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.3596111111111117e-05,
+      "loss": 0.0026,
+      "step": 69190
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.359518518518519e-05,
+      "loss": 0.0048,
+      "step": 69200
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.359425925925926e-05,
+      "loss": 0.0023,
+      "step": 69210
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.3593333333333335e-05,
+      "loss": 0.0031,
+      "step": 69220
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.359240740740741e-05,
+      "loss": 0.002,
+      "step": 69230
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.3591481481481485e-05,
+      "loss": 0.0022,
+      "step": 69240
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.359055555555555e-05,
+      "loss": 0.0034,
+      "step": 69250
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.358962962962963e-05,
+      "loss": 0.0023,
+      "step": 69260
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.3588703703703703e-05,
+      "loss": 0.0023,
+      "step": 69270
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.358777777777778e-05,
+      "loss": 0.0043,
+      "step": 69280
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.3586851851851854e-05,
+      "loss": 0.003,
+      "step": 69290
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.358592592592593e-05,
+      "loss": 0.0028,
+      "step": 69300
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.3585000000000004e-05,
+      "loss": 0.0028,
+      "step": 69310
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.358407407407408e-05,
+      "loss": 0.0037,
+      "step": 69320
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.3583148148148154e-05,
+      "loss": 0.003,
+      "step": 69330
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.358222222222222e-05,
+      "loss": 0.0053,
+      "step": 69340
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.35812962962963e-05,
+      "loss": 0.004,
+      "step": 69350
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.358037037037037e-05,
+      "loss": 0.0037,
+      "step": 69360
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.357944444444445e-05,
+      "loss": 0.0031,
+      "step": 69370
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.3578518518518516e-05,
+      "loss": 0.0045,
+      "step": 69380
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.357759259259259e-05,
+      "loss": 0.0022,
+      "step": 69390
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.357666666666667e-05,
+      "loss": 0.0051,
+      "step": 69400
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.357574074074075e-05,
+      "loss": 0.0042,
+      "step": 69410
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.3574814814814816e-05,
+      "loss": 0.0039,
+      "step": 69420
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.357388888888889e-05,
+      "loss": 0.0031,
+      "step": 69430
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.3572962962962966e-05,
+      "loss": 0.0041,
+      "step": 69440
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.357203703703704e-05,
+      "loss": 0.0022,
+      "step": 69450
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.357111111111111e-05,
+      "loss": 0.0055,
+      "step": 69460
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.3570185185185184e-05,
+      "loss": 0.0036,
+      "step": 69470
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.356925925925926e-05,
+      "loss": 0.0017,
+      "step": 69480
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.3568333333333334e-05,
+      "loss": 0.0028,
+      "step": 69490
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.356740740740741e-05,
+      "loss": 0.0042,
+      "step": 69500
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.3566481481481484e-05,
+      "loss": 0.0033,
+      "step": 69510
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.356555555555556e-05,
+      "loss": 0.0021,
+      "step": 69520
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.3564629629629635e-05,
+      "loss": 0.0034,
+      "step": 69530
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.356370370370371e-05,
+      "loss": 0.0028,
+      "step": 69540
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.356277777777778e-05,
+      "loss": 0.0037,
+      "step": 69550
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.356185185185185e-05,
+      "loss": 0.0025,
+      "step": 69560
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.356092592592593e-05,
+      "loss": 0.0012,
+      "step": 69570
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.356e-05,
+      "loss": 0.0026,
+      "step": 69580
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.355907407407407e-05,
+      "loss": 0.0026,
+      "step": 69590
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.3558148148148146e-05,
+      "loss": 0.0033,
+      "step": 69600
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.355722222222223e-05,
+      "loss": 0.0033,
+      "step": 69610
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.35562962962963e-05,
+      "loss": 0.0022,
+      "step": 69620
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.355537037037037e-05,
+      "loss": 0.0039,
+      "step": 69630
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.355444444444445e-05,
+      "loss": 0.0032,
+      "step": 69640
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.355351851851852e-05,
+      "loss": 0.0031,
+      "step": 69650
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.35525925925926e-05,
+      "loss": 0.0036,
+      "step": 69660
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.3551666666666665e-05,
+      "loss": 0.004,
+      "step": 69670
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.355074074074074e-05,
+      "loss": 0.0043,
+      "step": 69680
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.3549814814814815e-05,
+      "loss": 0.0014,
+      "step": 69690
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.354888888888889e-05,
+      "loss": 0.0028,
+      "step": 69700
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.3547962962962965e-05,
+      "loss": 0.0055,
+      "step": 69710
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.354703703703704e-05,
+      "loss": 0.0038,
+      "step": 69720
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.3546111111111115e-05,
+      "loss": 0.0038,
+      "step": 69730
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.354518518518519e-05,
+      "loss": 0.0049,
+      "step": 69740
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.3544259259259265e-05,
+      "loss": 0.0034,
+      "step": 69750
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.3543333333333334e-05,
+      "loss": 0.0017,
+      "step": 69760
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.354240740740741e-05,
+      "loss": 0.0054,
+      "step": 69770
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.3541481481481484e-05,
+      "loss": 0.0024,
+      "step": 69780
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.354055555555556e-05,
+      "loss": 0.0036,
+      "step": 69790
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.353962962962963e-05,
+      "loss": 0.0027,
+      "step": 69800
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.35387037037037e-05,
+      "loss": 0.0033,
+      "step": 69810
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.353777777777778e-05,
+      "loss": 0.0018,
+      "step": 69820
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.353685185185186e-05,
+      "loss": 0.0029,
+      "step": 69830
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.353592592592593e-05,
+      "loss": 0.0034,
+      "step": 69840
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.3535e-05,
+      "loss": 0.0028,
+      "step": 69850
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.353407407407408e-05,
+      "loss": 0.0042,
+      "step": 69860
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.353314814814815e-05,
+      "loss": 0.0048,
+      "step": 69870
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.353222222222222e-05,
+      "loss": 0.0043,
+      "step": 69880
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.3531296296296296e-05,
+      "loss": 0.0069,
+      "step": 69890
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.353037037037037e-05,
+      "loss": 0.0034,
+      "step": 69900
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.3529444444444446e-05,
+      "loss": 0.0059,
+      "step": 69910
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.352851851851852e-05,
+      "loss": 0.0092,
+      "step": 69920
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.352759259259259e-05,
+      "loss": 0.0043,
+      "step": 69930
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.352666666666667e-05,
+      "loss": 0.0021,
+      "step": 69940
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.3525740740740746e-05,
+      "loss": 0.0032,
+      "step": 69950
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.352481481481482e-05,
+      "loss": 0.0036,
+      "step": 69960
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.352388888888889e-05,
+      "loss": 0.0046,
+      "step": 69970
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.3522962962962965e-05,
+      "loss": 0.0048,
+      "step": 69980
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.352203703703704e-05,
+      "loss": 0.0036,
+      "step": 69990
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.3521111111111115e-05,
+      "loss": 0.0018,
+      "step": 70000
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.352018518518518e-05,
+      "loss": 0.0057,
+      "step": 70010
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.351925925925926e-05,
+      "loss": 0.0039,
+      "step": 70020
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.351833333333333e-05,
+      "loss": 0.004,
+      "step": 70030
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.351740740740741e-05,
+      "loss": 0.0018,
+      "step": 70040
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.351648148148148e-05,
+      "loss": 0.0043,
+      "step": 70050
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.351555555555556e-05,
+      "loss": 0.0025,
+      "step": 70060
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.351462962962963e-05,
+      "loss": 0.0029,
+      "step": 70070
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.351370370370371e-05,
+      "loss": 0.0029,
+      "step": 70080
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.3512777777777783e-05,
+      "loss": 0.0036,
+      "step": 70090
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.351185185185185e-05,
+      "loss": 0.0043,
+      "step": 70100
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.351092592592593e-05,
+      "loss": 0.0031,
+      "step": 70110
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.351e-05,
+      "loss": 0.0047,
+      "step": 70120
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.350907407407408e-05,
+      "loss": 0.0062,
+      "step": 70130
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.3508148148148145e-05,
+      "loss": 0.0029,
+      "step": 70140
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.350722222222223e-05,
+      "loss": 0.0042,
+      "step": 70150
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.35062962962963e-05,
+      "loss": 0.0052,
+      "step": 70160
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.350537037037038e-05,
+      "loss": 0.0023,
+      "step": 70170
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.3504444444444445e-05,
+      "loss": 0.0033,
+      "step": 70180
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.350351851851852e-05,
+      "loss": 0.0025,
+      "step": 70190
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.3502592592592596e-05,
+      "loss": 0.0037,
+      "step": 70200
+    },
+    {
+      "epoch": 13.0,
+      "eval_loss": 0.016600701957941055,
+      "eval_runtime": 153.5709,
+      "eval_samples_per_second": 25.981,
+      "eval_steps_per_second": 3.249,
+      "step": 70200
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.350166666666667e-05,
+      "loss": 0.0022,
+      "step": 70210
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.350074074074074e-05,
+      "loss": 0.0023,
+      "step": 70220
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.3499814814814814e-05,
+      "loss": 0.0041,
+      "step": 70230
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.349888888888889e-05,
+      "loss": 0.0019,
+      "step": 70240
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.3497962962962964e-05,
+      "loss": 0.0029,
+      "step": 70250
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.349703703703704e-05,
+      "loss": 0.0024,
+      "step": 70260
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.3496111111111114e-05,
+      "loss": 0.002,
+      "step": 70270
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.349527777777778e-05,
+      "loss": 0.0032,
+      "step": 70280
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.3494351851851854e-05,
+      "loss": 0.0057,
+      "step": 70290
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.349342592592593e-05,
+      "loss": 0.0019,
+      "step": 70300
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.3492500000000004e-05,
+      "loss": 0.004,
+      "step": 70310
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.349157407407408e-05,
+      "loss": 0.0014,
+      "step": 70320
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.349064814814815e-05,
+      "loss": 0.0029,
+      "step": 70330
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.348972222222222e-05,
+      "loss": 0.0039,
+      "step": 70340
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.34887962962963e-05,
+      "loss": 0.0021,
+      "step": 70350
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.348787037037037e-05,
+      "loss": 0.0062,
+      "step": 70360
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.348694444444444e-05,
+      "loss": 0.0018,
+      "step": 70370
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.348601851851852e-05,
+      "loss": 0.0025,
+      "step": 70380
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.34850925925926e-05,
+      "loss": 0.0042,
+      "step": 70390
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.348416666666667e-05,
+      "loss": 0.0036,
+      "step": 70400
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.348324074074074e-05,
+      "loss": 0.0043,
+      "step": 70410
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.3482314814814816e-05,
+      "loss": 0.0033,
+      "step": 70420
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.348138888888889e-05,
+      "loss": 0.0072,
+      "step": 70430
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.3480462962962966e-05,
+      "loss": 0.0027,
+      "step": 70440
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.347953703703704e-05,
+      "loss": 0.0043,
+      "step": 70450
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.347861111111111e-05,
+      "loss": 0.0032,
+      "step": 70460
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.3477685185185185e-05,
+      "loss": 0.0026,
+      "step": 70470
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.347675925925926e-05,
+      "loss": 0.0036,
+      "step": 70480
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.3475833333333335e-05,
+      "loss": 0.0047,
+      "step": 70490
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.347490740740741e-05,
+      "loss": 0.004,
+      "step": 70500
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.3473981481481485e-05,
+      "loss": 0.0026,
+      "step": 70510
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.347305555555556e-05,
+      "loss": 0.0035,
+      "step": 70520
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.3472129629629635e-05,
+      "loss": 0.0033,
+      "step": 70530
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.34712037037037e-05,
+      "loss": 0.0034,
+      "step": 70540
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.347027777777778e-05,
+      "loss": 0.006,
+      "step": 70550
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.3469351851851853e-05,
+      "loss": 0.0052,
+      "step": 70560
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.346842592592593e-05,
+      "loss": 0.0036,
+      "step": 70570
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.34675e-05,
+      "loss": 0.0015,
+      "step": 70580
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.346657407407407e-05,
+      "loss": 0.0034,
+      "step": 70590
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.3465648148148154e-05,
+      "loss": 0.0036,
+      "step": 70600
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.346472222222223e-05,
+      "loss": 0.0044,
+      "step": 70610
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.34637962962963e-05,
+      "loss": 0.002,
+      "step": 70620
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.346287037037037e-05,
+      "loss": 0.0015,
+      "step": 70630
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.346194444444445e-05,
+      "loss": 0.0023,
+      "step": 70640
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.346101851851852e-05,
+      "loss": 0.0031,
+      "step": 70650
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.34600925925926e-05,
+      "loss": 0.0029,
+      "step": 70660
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.3459166666666665e-05,
+      "loss": 0.0046,
+      "step": 70670
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.345824074074074e-05,
+      "loss": 0.0026,
+      "step": 70680
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.3457314814814816e-05,
+      "loss": 0.0012,
+      "step": 70690
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.345638888888889e-05,
+      "loss": 0.003,
+      "step": 70700
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.3455462962962966e-05,
+      "loss": 0.001,
+      "step": 70710
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.345453703703704e-05,
+      "loss": 0.0026,
+      "step": 70720
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.3453611111111116e-05,
+      "loss": 0.0027,
+      "step": 70730
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.345268518518519e-05,
+      "loss": 0.0031,
+      "step": 70740
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.345175925925926e-05,
+      "loss": 0.0021,
+      "step": 70750
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.3450833333333334e-05,
+      "loss": 0.0043,
+      "step": 70760
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.344990740740741e-05,
+      "loss": 0.0031,
+      "step": 70770
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.3448981481481484e-05,
+      "loss": 0.003,
+      "step": 70780
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.344805555555555e-05,
+      "loss": 0.0033,
+      "step": 70790
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.344712962962963e-05,
+      "loss": 0.0041,
+      "step": 70800
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.344620370370371e-05,
+      "loss": 0.0032,
+      "step": 70810
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.3445277777777785e-05,
+      "loss": 0.0016,
+      "step": 70820
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.344435185185185e-05,
+      "loss": 0.0064,
+      "step": 70830
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.344342592592593e-05,
+      "loss": 0.0032,
+      "step": 70840
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.34425e-05,
+      "loss": 0.0037,
+      "step": 70850
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.344157407407408e-05,
+      "loss": 0.0023,
+      "step": 70860
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.344064814814815e-05,
+      "loss": 0.003,
+      "step": 70870
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.343972222222222e-05,
+      "loss": 0.0022,
+      "step": 70880
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.3438796296296296e-05,
+      "loss": 0.0036,
+      "step": 70890
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.343787037037037e-05,
+      "loss": 0.0028,
+      "step": 70900
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.3436944444444446e-05,
+      "loss": 0.0021,
+      "step": 70910
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.343601851851852e-05,
+      "loss": 0.0023,
+      "step": 70920
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.3435092592592597e-05,
+      "loss": 0.0025,
+      "step": 70930
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.343416666666667e-05,
+      "loss": 0.004,
+      "step": 70940
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.343324074074075e-05,
+      "loss": 0.0067,
+      "step": 70950
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.3432314814814815e-05,
+      "loss": 0.0029,
+      "step": 70960
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.343138888888889e-05,
+      "loss": 0.0026,
+      "step": 70970
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.3430462962962965e-05,
+      "loss": 0.002,
+      "step": 70980
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.342953703703704e-05,
+      "loss": 0.0033,
+      "step": 70990
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.342861111111111e-05,
+      "loss": 0.0028,
+      "step": 71000
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.3427685185185183e-05,
+      "loss": 0.0033,
+      "step": 71010
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.342675925925926e-05,
+      "loss": 0.0034,
+      "step": 71020
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.342583333333334e-05,
+      "loss": 0.0018,
+      "step": 71030
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.342490740740741e-05,
+      "loss": 0.0018,
+      "step": 71040
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.3423981481481484e-05,
+      "loss": 0.0039,
+      "step": 71050
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.342305555555556e-05,
+      "loss": 0.0043,
+      "step": 71060
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.3422129629629634e-05,
+      "loss": 0.0019,
+      "step": 71070
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.342120370370371e-05,
+      "loss": 0.0023,
+      "step": 71080
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.342027777777778e-05,
+      "loss": 0.0022,
+      "step": 71090
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.341935185185185e-05,
+      "loss": 0.0019,
+      "step": 71100
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.341842592592593e-05,
+      "loss": 0.0018,
+      "step": 71110
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.34175e-05,
+      "loss": 0.0032,
+      "step": 71120
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.341657407407407e-05,
+      "loss": 0.0034,
+      "step": 71130
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.341564814814815e-05,
+      "loss": 0.002,
+      "step": 71140
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.341472222222223e-05,
+      "loss": 0.005,
+      "step": 71150
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.34137962962963e-05,
+      "loss": 0.0054,
+      "step": 71160
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.341287037037037e-05,
+      "loss": 0.0068,
+      "step": 71170
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.3411944444444446e-05,
+      "loss": 0.0044,
+      "step": 71180
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.341101851851852e-05,
+      "loss": 0.0016,
+      "step": 71190
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.3410092592592596e-05,
+      "loss": 0.0034,
+      "step": 71200
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.3409166666666664e-05,
+      "loss": 0.0031,
+      "step": 71210
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.3408333333333336e-05,
+      "loss": 0.0041,
+      "step": 71220
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.340740740740741e-05,
+      "loss": 0.0016,
+      "step": 71230
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.340648148148148e-05,
+      "loss": 0.0017,
+      "step": 71240
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.3405555555555554e-05,
+      "loss": 0.0013,
+      "step": 71250
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.3404629629629636e-05,
+      "loss": 0.0031,
+      "step": 71260
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.3403703703703704e-05,
+      "loss": 0.0027,
+      "step": 71270
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.340277777777778e-05,
+      "loss": 0.0038,
+      "step": 71280
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.3401851851851854e-05,
+      "loss": 0.0015,
+      "step": 71290
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.340092592592593e-05,
+      "loss": 0.0039,
+      "step": 71300
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.3400000000000005e-05,
+      "loss": 0.0027,
+      "step": 71310
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.339907407407407e-05,
+      "loss": 0.0029,
+      "step": 71320
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.339814814814815e-05,
+      "loss": 0.0043,
+      "step": 71330
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.339722222222222e-05,
+      "loss": 0.0023,
+      "step": 71340
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.33962962962963e-05,
+      "loss": 0.0032,
+      "step": 71350
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.339537037037037e-05,
+      "loss": 0.0015,
+      "step": 71360
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.339444444444445e-05,
+      "loss": 0.002,
+      "step": 71370
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.339351851851852e-05,
+      "loss": 0.003,
+      "step": 71380
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.33925925925926e-05,
+      "loss": 0.0036,
+      "step": 71390
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.3391666666666667e-05,
+      "loss": 0.0014,
+      "step": 71400
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.339074074074074e-05,
+      "loss": 0.0041,
+      "step": 71410
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.338981481481482e-05,
+      "loss": 0.0031,
+      "step": 71420
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.338888888888889e-05,
+      "loss": 0.0048,
+      "step": 71430
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.338796296296297e-05,
+      "loss": 0.0033,
+      "step": 71440
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.3387037037037035e-05,
+      "loss": 0.0035,
+      "step": 71450
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.338611111111111e-05,
+      "loss": 0.0023,
+      "step": 71460
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.338518518518519e-05,
+      "loss": 0.0019,
+      "step": 71470
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.338425925925926e-05,
+      "loss": 0.0032,
+      "step": 71480
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.3383333333333335e-05,
+      "loss": 0.0021,
+      "step": 71490
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.338240740740741e-05,
+      "loss": 0.0023,
+      "step": 71500
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.3381481481481485e-05,
+      "loss": 0.0023,
+      "step": 71510
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.338055555555556e-05,
+      "loss": 0.0016,
+      "step": 71520
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.337962962962963e-05,
+      "loss": 0.0027,
+      "step": 71530
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.3378703703703704e-05,
+      "loss": 0.0019,
+      "step": 71540
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.337777777777778e-05,
+      "loss": 0.0021,
+      "step": 71550
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.3376851851851854e-05,
+      "loss": 0.0016,
+      "step": 71560
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.337592592592592e-05,
+      "loss": 0.0035,
+      "step": 71570
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.3375000000000004e-05,
+      "loss": 0.002,
+      "step": 71580
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.337407407407408e-05,
+      "loss": 0.0029,
+      "step": 71590
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.3373148148148154e-05,
+      "loss": 0.0047,
+      "step": 71600
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.337222222222222e-05,
+      "loss": 0.003,
+      "step": 71610
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.33712962962963e-05,
+      "loss": 0.0024,
+      "step": 71620
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.337037037037037e-05,
+      "loss": 0.003,
+      "step": 71630
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.336944444444445e-05,
+      "loss": 0.0029,
+      "step": 71640
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.336851851851852e-05,
+      "loss": 0.0021,
+      "step": 71650
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.336759259259259e-05,
+      "loss": 0.0027,
+      "step": 71660
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.3366666666666666e-05,
+      "loss": 0.0017,
+      "step": 71670
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.336574074074074e-05,
+      "loss": 0.0027,
+      "step": 71680
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.3364814814814816e-05,
+      "loss": 0.0016,
+      "step": 71690
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.336388888888889e-05,
+      "loss": 0.0045,
+      "step": 71700
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.3362962962962966e-05,
+      "loss": 0.0013,
+      "step": 71710
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.336203703703704e-05,
+      "loss": 0.0044,
+      "step": 71720
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.3361111111111116e-05,
+      "loss": 0.0028,
+      "step": 71730
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.3360185185185185e-05,
+      "loss": 0.0015,
+      "step": 71740
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.335925925925926e-05,
+      "loss": 0.0034,
+      "step": 71750
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.3358333333333335e-05,
+      "loss": 0.0027,
+      "step": 71760
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.335740740740741e-05,
+      "loss": 0.0037,
+      "step": 71770
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.335648148148148e-05,
+      "loss": 0.0033,
+      "step": 71780
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.335555555555556e-05,
+      "loss": 0.0034,
+      "step": 71790
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.3354629629629635e-05,
+      "loss": 0.0042,
+      "step": 71800
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.335370370370371e-05,
+      "loss": 0.0021,
+      "step": 71810
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.335277777777778e-05,
+      "loss": 0.0025,
+      "step": 71820
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.335185185185185e-05,
+      "loss": 0.0055,
+      "step": 71830
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.335092592592593e-05,
+      "loss": 0.0044,
+      "step": 71840
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.335e-05,
+      "loss": 0.0026,
+      "step": 71850
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.334907407407408e-05,
+      "loss": 0.0019,
+      "step": 71860
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.334814814814815e-05,
+      "loss": 0.0069,
+      "step": 71870
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.334722222222222e-05,
+      "loss": 0.0039,
+      "step": 71880
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.33462962962963e-05,
+      "loss": 0.002,
+      "step": 71890
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.334537037037037e-05,
+      "loss": 0.0028,
+      "step": 71900
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.334444444444445e-05,
+      "loss": 0.0024,
+      "step": 71910
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.334351851851852e-05,
+      "loss": 0.0022,
+      "step": 71920
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.33425925925926e-05,
+      "loss": 0.001,
+      "step": 71930
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.334166666666667e-05,
+      "loss": 0.0016,
+      "step": 71940
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.334074074074074e-05,
+      "loss": 0.0016,
+      "step": 71950
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.3339814814814815e-05,
+      "loss": 0.004,
+      "step": 71960
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.333888888888889e-05,
+      "loss": 0.0036,
+      "step": 71970
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.3337962962962966e-05,
+      "loss": 0.0016,
+      "step": 71980
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.3337037037037034e-05,
+      "loss": 0.0023,
+      "step": 71990
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.333611111111111e-05,
+      "loss": 0.0016,
+      "step": 72000
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.333518518518519e-05,
+      "loss": 0.0026,
+      "step": 72010
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.3334259259259266e-05,
+      "loss": 0.0022,
+      "step": 72020
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.3333333333333334e-05,
+      "loss": 0.0029,
+      "step": 72030
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.333240740740741e-05,
+      "loss": 0.0037,
+      "step": 72040
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.3331481481481484e-05,
+      "loss": 0.0024,
+      "step": 72050
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.333055555555556e-05,
+      "loss": 0.0033,
+      "step": 72060
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.3329629629629634e-05,
+      "loss": 0.0042,
+      "step": 72070
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.33287037037037e-05,
+      "loss": 0.0036,
+      "step": 72080
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.332777777777778e-05,
+      "loss": 0.0024,
+      "step": 72090
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.332685185185185e-05,
+      "loss": 0.0025,
+      "step": 72100
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.332592592592593e-05,
+      "loss": 0.0027,
+      "step": 72110
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.3325e-05,
+      "loss": 0.0032,
+      "step": 72120
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.332407407407408e-05,
+      "loss": 0.0045,
+      "step": 72130
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.332314814814815e-05,
+      "loss": 0.0017,
+      "step": 72140
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.332222222222223e-05,
+      "loss": 0.0025,
+      "step": 72150
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.3321296296296296e-05,
+      "loss": 0.0023,
+      "step": 72160
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.332037037037037e-05,
+      "loss": 0.0021,
+      "step": 72170
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.3319444444444446e-05,
+      "loss": 0.0037,
+      "step": 72180
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.331851851851852e-05,
+      "loss": 0.0041,
+      "step": 72190
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.331759259259259e-05,
+      "loss": 0.005,
+      "step": 72200
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.3316666666666665e-05,
+      "loss": 0.0017,
+      "step": 72210
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.331574074074074e-05,
+      "loss": 0.0024,
+      "step": 72220
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.331481481481482e-05,
+      "loss": 0.0013,
+      "step": 72230
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.331388888888889e-05,
+      "loss": 0.0023,
+      "step": 72240
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.3312962962962965e-05,
+      "loss": 0.0032,
+      "step": 72250
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.331203703703704e-05,
+      "loss": 0.0018,
+      "step": 72260
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.3311111111111115e-05,
+      "loss": 0.0027,
+      "step": 72270
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.331018518518519e-05,
+      "loss": 0.0029,
+      "step": 72280
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.330925925925926e-05,
+      "loss": 0.0021,
+      "step": 72290
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.3308333333333333e-05,
+      "loss": 0.0028,
+      "step": 72300
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.330740740740741e-05,
+      "loss": 0.0032,
+      "step": 72310
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.3306481481481484e-05,
+      "loss": 0.0034,
+      "step": 72320
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.330555555555556e-05,
+      "loss": 0.0032,
+      "step": 72330
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.3304629629629634e-05,
+      "loss": 0.0046,
+      "step": 72340
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.330370370370371e-05,
+      "loss": 0.0027,
+      "step": 72350
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.3302777777777784e-05,
+      "loss": 0.0036,
+      "step": 72360
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.330185185185185e-05,
+      "loss": 0.0014,
+      "step": 72370
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.330092592592593e-05,
+      "loss": 0.0025,
+      "step": 72380
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.33e-05,
+      "loss": 0.0033,
+      "step": 72390
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.329907407407408e-05,
+      "loss": 0.0037,
+      "step": 72400
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.3298148148148146e-05,
+      "loss": 0.0041,
+      "step": 72410
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.329722222222222e-05,
+      "loss": 0.0027,
+      "step": 72420
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.3296296296296296e-05,
+      "loss": 0.0035,
+      "step": 72430
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.329537037037038e-05,
+      "loss": 0.0026,
+      "step": 72440
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.3294444444444446e-05,
+      "loss": 0.0013,
+      "step": 72450
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.329351851851852e-05,
+      "loss": 0.0029,
+      "step": 72460
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.3292592592592596e-05,
+      "loss": 0.003,
+      "step": 72470
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.329166666666667e-05,
+      "loss": 0.0038,
+      "step": 72480
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.3290740740740746e-05,
+      "loss": 0.0012,
+      "step": 72490
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.3289814814814814e-05,
+      "loss": 0.0016,
+      "step": 72500
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.328888888888889e-05,
+      "loss": 0.004,
+      "step": 72510
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.3287962962962964e-05,
+      "loss": 0.0031,
+      "step": 72520
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.328703703703704e-05,
+      "loss": 0.0034,
+      "step": 72530
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.328611111111111e-05,
+      "loss": 0.0034,
+      "step": 72540
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.328518518518519e-05,
+      "loss": 0.0017,
+      "step": 72550
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.3284259259259265e-05,
+      "loss": 0.0026,
+      "step": 72560
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.328333333333334e-05,
+      "loss": 0.0026,
+      "step": 72570
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.328240740740741e-05,
+      "loss": 0.0036,
+      "step": 72580
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.328148148148148e-05,
+      "loss": 0.0029,
+      "step": 72590
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.328055555555556e-05,
+      "loss": 0.002,
+      "step": 72600
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.327962962962963e-05,
+      "loss": 0.0016,
+      "step": 72610
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.32787037037037e-05,
+      "loss": 0.0016,
+      "step": 72620
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.3277777777777776e-05,
+      "loss": 0.0022,
+      "step": 72630
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.327685185185185e-05,
+      "loss": 0.0047,
+      "step": 72640
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.3275925925925927e-05,
+      "loss": 0.0036,
+      "step": 72650
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.3275e-05,
+      "loss": 0.005,
+      "step": 72660
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.327407407407408e-05,
+      "loss": 0.0039,
+      "step": 72670
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.327314814814815e-05,
+      "loss": 0.0023,
+      "step": 72680
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.327222222222223e-05,
+      "loss": 0.0041,
+      "step": 72690
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.32712962962963e-05,
+      "loss": 0.0035,
+      "step": 72700
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.327037037037037e-05,
+      "loss": 0.0026,
+      "step": 72710
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.3269444444444445e-05,
+      "loss": 0.0031,
+      "step": 72720
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.326851851851852e-05,
+      "loss": 0.0041,
+      "step": 72730
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.3267592592592595e-05,
+      "loss": 0.003,
+      "step": 72740
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.3266666666666664e-05,
+      "loss": 0.0042,
+      "step": 72750
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.3265740740740745e-05,
+      "loss": 0.0022,
+      "step": 72760
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.326481481481482e-05,
+      "loss": 0.0013,
+      "step": 72770
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.3263888888888895e-05,
+      "loss": 0.0022,
+      "step": 72780
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.3262962962962964e-05,
+      "loss": 0.0044,
+      "step": 72790
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.326203703703704e-05,
+      "loss": 0.0019,
+      "step": 72800
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.3261111111111114e-05,
+      "loss": 0.002,
+      "step": 72810
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.326018518518519e-05,
+      "loss": 0.0032,
+      "step": 72820
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.325925925925926e-05,
+      "loss": 0.0022,
+      "step": 72830
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.325833333333333e-05,
+      "loss": 0.0046,
+      "step": 72840
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.325740740740741e-05,
+      "loss": 0.003,
+      "step": 72850
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.325648148148148e-05,
+      "loss": 0.0016,
+      "step": 72860
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.325555555555556e-05,
+      "loss": 0.0024,
+      "step": 72870
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.325462962962963e-05,
+      "loss": 0.0031,
+      "step": 72880
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.325370370370371e-05,
+      "loss": 0.0054,
+      "step": 72890
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.325277777777778e-05,
+      "loss": 0.0032,
+      "step": 72900
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.325185185185186e-05,
+      "loss": 0.0049,
+      "step": 72910
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.3250925925925926e-05,
+      "loss": 0.0065,
+      "step": 72920
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.325e-05,
+      "loss": 0.0025,
+      "step": 72930
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.3249074074074076e-05,
+      "loss": 0.004,
+      "step": 72940
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.324814814814815e-05,
+      "loss": 0.0025,
+      "step": 72950
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.324722222222222e-05,
+      "loss": 0.0035,
+      "step": 72960
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.3246296296296294e-05,
+      "loss": 0.002,
+      "step": 72970
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.3245370370370376e-05,
+      "loss": 0.0082,
+      "step": 72980
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.324444444444445e-05,
+      "loss": 0.0026,
+      "step": 72990
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.324351851851852e-05,
+      "loss": 0.0048,
+      "step": 73000
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.3242592592592595e-05,
+      "loss": 0.0025,
+      "step": 73010
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.324166666666667e-05,
+      "loss": 0.003,
+      "step": 73020
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.3240740740740745e-05,
+      "loss": 0.0049,
+      "step": 73030
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.323981481481481e-05,
+      "loss": 0.0021,
+      "step": 73040
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.323888888888889e-05,
+      "loss": 0.0027,
+      "step": 73050
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.323796296296296e-05,
+      "loss": 0.0021,
+      "step": 73060
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.323703703703704e-05,
+      "loss": 0.0029,
+      "step": 73070
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.323611111111111e-05,
+      "loss": 0.0029,
+      "step": 73080
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.323518518518519e-05,
+      "loss": 0.0026,
+      "step": 73090
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.323425925925926e-05,
+      "loss": 0.0026,
+      "step": 73100
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.323333333333334e-05,
+      "loss": 0.0031,
+      "step": 73110
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.3232407407407413e-05,
+      "loss": 0.002,
+      "step": 73120
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.323148148148148e-05,
+      "loss": 0.0023,
+      "step": 73130
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.323055555555556e-05,
+      "loss": 0.0014,
+      "step": 73140
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.322962962962963e-05,
+      "loss": 0.0019,
+      "step": 73150
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.322870370370371e-05,
+      "loss": 0.0015,
+      "step": 73160
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.3227777777777775e-05,
+      "loss": 0.0022,
+      "step": 73170
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.322685185185185e-05,
+      "loss": 0.0029,
+      "step": 73180
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.3225925925925925e-05,
+      "loss": 0.0027,
+      "step": 73190
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.322500000000001e-05,
+      "loss": 0.0019,
+      "step": 73200
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.3224074074074075e-05,
+      "loss": 0.0017,
+      "step": 73210
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.322314814814815e-05,
+      "loss": 0.0023,
+      "step": 73220
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.3222222222222226e-05,
+      "loss": 0.0017,
+      "step": 73230
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.32212962962963e-05,
+      "loss": 0.0013,
+      "step": 73240
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.322037037037037e-05,
+      "loss": 0.0014,
+      "step": 73250
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.3219444444444444e-05,
+      "loss": 0.0031,
+      "step": 73260
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.321851851851852e-05,
+      "loss": 0.0018,
+      "step": 73270
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.3217592592592594e-05,
+      "loss": 0.0048,
+      "step": 73280
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.321666666666667e-05,
+      "loss": 0.003,
+      "step": 73290
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.3215740740740744e-05,
+      "loss": 0.0023,
+      "step": 73300
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.321481481481482e-05,
+      "loss": 0.0032,
+      "step": 73310
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.3213888888888894e-05,
+      "loss": 0.0037,
+      "step": 73320
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.321296296296297e-05,
+      "loss": 0.004,
+      "step": 73330
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.321203703703704e-05,
+      "loss": 0.0033,
+      "step": 73340
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.321111111111111e-05,
+      "loss": 0.0028,
+      "step": 73350
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.321018518518519e-05,
+      "loss": 0.0034,
+      "step": 73360
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.320925925925926e-05,
+      "loss": 0.0037,
+      "step": 73370
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.320833333333333e-05,
+      "loss": 0.0033,
+      "step": 73380
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.3207407407407406e-05,
+      "loss": 0.0024,
+      "step": 73390
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.320648148148148e-05,
+      "loss": 0.0045,
+      "step": 73400
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.320555555555556e-05,
+      "loss": 0.0036,
+      "step": 73410
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.320462962962963e-05,
+      "loss": 0.0031,
+      "step": 73420
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.3203703703703706e-05,
+      "loss": 0.0029,
+      "step": 73430
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.320277777777778e-05,
+      "loss": 0.0023,
+      "step": 73440
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.3201851851851856e-05,
+      "loss": 0.0024,
+      "step": 73450
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.3200925925925925e-05,
+      "loss": 0.0027,
+      "step": 73460
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.32e-05,
+      "loss": 0.0035,
+      "step": 73470
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.3199074074074075e-05,
+      "loss": 0.0036,
+      "step": 73480
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.319814814814815e-05,
+      "loss": 0.0016,
+      "step": 73490
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.3197222222222225e-05,
+      "loss": 0.0018,
+      "step": 73500
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.319629629629629e-05,
+      "loss": 0.0013,
+      "step": 73510
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.3195370370370375e-05,
+      "loss": 0.0024,
+      "step": 73520
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.319444444444445e-05,
+      "loss": 0.0024,
+      "step": 73530
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.3193518518518525e-05,
+      "loss": 0.0038,
+      "step": 73540
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.3192592592592593e-05,
+      "loss": 0.0028,
+      "step": 73550
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.319166666666667e-05,
+      "loss": 0.0024,
+      "step": 73560
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.3190740740740744e-05,
+      "loss": 0.0043,
+      "step": 73570
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.318981481481482e-05,
+      "loss": 0.005,
+      "step": 73580
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.318888888888889e-05,
+      "loss": 0.0032,
+      "step": 73590
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.318796296296296e-05,
+      "loss": 0.0029,
+      "step": 73600
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.318703703703704e-05,
+      "loss": 0.0047,
+      "step": 73610
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.318611111111111e-05,
+      "loss": 0.0035,
+      "step": 73620
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.318518518518519e-05,
+      "loss": 0.0067,
+      "step": 73630
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.318425925925926e-05,
+      "loss": 0.0055,
+      "step": 73640
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.318333333333334e-05,
+      "loss": 0.0052,
+      "step": 73650
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.318240740740741e-05,
+      "loss": 0.0044,
+      "step": 73660
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.318148148148148e-05,
+      "loss": 0.0033,
+      "step": 73670
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.3180555555555556e-05,
+      "loss": 0.0037,
+      "step": 73680
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.317962962962963e-05,
+      "loss": 0.0028,
+      "step": 73690
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.3178703703703706e-05,
+      "loss": 0.0023,
+      "step": 73700
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.317777777777778e-05,
+      "loss": 0.0051,
+      "step": 73710
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.317685185185185e-05,
+      "loss": 0.006,
+      "step": 73720
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.3175925925925924e-05,
+      "loss": 0.0027,
+      "step": 73730
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.3175000000000006e-05,
+      "loss": 0.0035,
+      "step": 73740
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.317407407407408e-05,
+      "loss": 0.005,
+      "step": 73750
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.317314814814815e-05,
+      "loss": 0.0046,
+      "step": 73760
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.3172222222222224e-05,
+      "loss": 0.0032,
+      "step": 73770
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.31712962962963e-05,
+      "loss": 0.0057,
+      "step": 73780
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.3170370370370374e-05,
+      "loss": 0.004,
+      "step": 73790
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.316944444444444e-05,
+      "loss": 0.0047,
+      "step": 73800
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.316851851851852e-05,
+      "loss": 0.0026,
+      "step": 73810
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.316759259259259e-05,
+      "loss": 0.0021,
+      "step": 73820
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.316666666666667e-05,
+      "loss": 0.0032,
+      "step": 73830
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.316574074074074e-05,
+      "loss": 0.0028,
+      "step": 73840
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.316481481481482e-05,
+      "loss": 0.0044,
+      "step": 73850
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.316388888888889e-05,
+      "loss": 0.002,
+      "step": 73860
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.316296296296297e-05,
+      "loss": 0.0024,
+      "step": 73870
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.3162037037037036e-05,
+      "loss": 0.0037,
+      "step": 73880
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.316111111111111e-05,
+      "loss": 0.0034,
+      "step": 73890
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.3160185185185186e-05,
+      "loss": 0.0035,
+      "step": 73900
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.315925925925926e-05,
+      "loss": 0.0036,
+      "step": 73910
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.3158333333333337e-05,
+      "loss": 0.0031,
+      "step": 73920
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.3157407407407405e-05,
+      "loss": 0.0025,
+      "step": 73930
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.315648148148148e-05,
+      "loss": 0.0034,
+      "step": 73940
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.315555555555556e-05,
+      "loss": 0.0027,
+      "step": 73950
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.315462962962964e-05,
+      "loss": 0.0033,
+      "step": 73960
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.3153703703703705e-05,
+      "loss": 0.0012,
+      "step": 73970
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.315277777777778e-05,
+      "loss": 0.0025,
+      "step": 73980
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.3151851851851855e-05,
+      "loss": 0.0038,
+      "step": 73990
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.315092592592593e-05,
+      "loss": 0.005,
+      "step": 74000
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.315e-05,
+      "loss": 0.0048,
+      "step": 74010
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.3149074074074074e-05,
+      "loss": 0.0067,
+      "step": 74020
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.314814814814815e-05,
+      "loss": 0.0029,
+      "step": 74030
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.3147222222222224e-05,
+      "loss": 0.0026,
+      "step": 74040
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.31462962962963e-05,
+      "loss": 0.0031,
+      "step": 74050
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.3145370370370374e-05,
+      "loss": 0.004,
+      "step": 74060
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.314444444444445e-05,
+      "loss": 0.0022,
+      "step": 74070
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.3143518518518524e-05,
+      "loss": 0.0032,
+      "step": 74080
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.314259259259259e-05,
+      "loss": 0.0037,
+      "step": 74090
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.314166666666667e-05,
+      "loss": 0.0023,
+      "step": 74100
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.314074074074074e-05,
+      "loss": 0.0021,
+      "step": 74110
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.313981481481482e-05,
+      "loss": 0.0029,
+      "step": 74120
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.313888888888889e-05,
+      "loss": 0.0029,
+      "step": 74130
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.313796296296296e-05,
+      "loss": 0.0022,
+      "step": 74140
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.3137037037037036e-05,
+      "loss": 0.0036,
+      "step": 74150
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.313611111111111e-05,
+      "loss": 0.0022,
+      "step": 74160
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.313518518518519e-05,
+      "loss": 0.0022,
+      "step": 74170
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.313425925925926e-05,
+      "loss": 0.0021,
+      "step": 74180
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.3133333333333336e-05,
+      "loss": 0.0032,
+      "step": 74190
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.313240740740741e-05,
+      "loss": 0.0042,
+      "step": 74200
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.3131481481481486e-05,
+      "loss": 0.0033,
+      "step": 74210
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.3130555555555554e-05,
+      "loss": 0.0045,
+      "step": 74220
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.312962962962963e-05,
+      "loss": 0.004,
+      "step": 74230
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.3128703703703704e-05,
+      "loss": 0.0025,
+      "step": 74240
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.312777777777778e-05,
+      "loss": 0.005,
+      "step": 74250
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.3126851851851855e-05,
+      "loss": 0.0052,
+      "step": 74260
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.312592592592593e-05,
+      "loss": 0.002,
+      "step": 74270
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.3125000000000005e-05,
+      "loss": 0.0018,
+      "step": 74280
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.312407407407408e-05,
+      "loss": 0.0044,
+      "step": 74290
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.312314814814815e-05,
+      "loss": 0.0029,
+      "step": 74300
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.312222222222222e-05,
+      "loss": 0.0036,
+      "step": 74310
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.31212962962963e-05,
+      "loss": 0.0022,
+      "step": 74320
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.312037037037037e-05,
+      "loss": 0.0028,
+      "step": 74330
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.311944444444445e-05,
+      "loss": 0.0023,
+      "step": 74340
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.3118518518518517e-05,
+      "loss": 0.0029,
+      "step": 74350
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.311759259259259e-05,
+      "loss": 0.0031,
+      "step": 74360
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.311666666666667e-05,
+      "loss": 0.0067,
+      "step": 74370
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.311574074074075e-05,
+      "loss": 0.0039,
+      "step": 74380
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.311481481481482e-05,
+      "loss": 0.0054,
+      "step": 74390
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.311388888888889e-05,
+      "loss": 0.003,
+      "step": 74400
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.311296296296297e-05,
+      "loss": 0.0039,
+      "step": 74410
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.311203703703704e-05,
+      "loss": 0.0043,
+      "step": 74420
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.311111111111111e-05,
+      "loss": 0.0017,
+      "step": 74430
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.3110185185185185e-05,
+      "loss": 0.0023,
+      "step": 74440
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.310925925925926e-05,
+      "loss": 0.0022,
+      "step": 74450
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.3108333333333335e-05,
+      "loss": 0.0025,
+      "step": 74460
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.310740740740741e-05,
+      "loss": 0.004,
+      "step": 74470
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.310648148148148e-05,
+      "loss": 0.0023,
+      "step": 74480
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.310555555555556e-05,
+      "loss": 0.0023,
+      "step": 74490
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.3104629629629636e-05,
+      "loss": 0.0028,
+      "step": 74500
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.310370370370371e-05,
+      "loss": 0.0019,
+      "step": 74510
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.310277777777778e-05,
+      "loss": 0.002,
+      "step": 74520
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.3101851851851854e-05,
+      "loss": 0.0024,
+      "step": 74530
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.310092592592593e-05,
+      "loss": 0.0026,
+      "step": 74540
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.3100000000000004e-05,
+      "loss": 0.0027,
+      "step": 74550
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.309907407407407e-05,
+      "loss": 0.0023,
+      "step": 74560
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.309814814814815e-05,
+      "loss": 0.0032,
+      "step": 74570
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.309722222222222e-05,
+      "loss": 0.004,
+      "step": 74580
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.30962962962963e-05,
+      "loss": 0.0026,
+      "step": 74590
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.309537037037037e-05,
+      "loss": 0.0059,
+      "step": 74600
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.309444444444445e-05,
+      "loss": 0.0025,
+      "step": 74610
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.309351851851852e-05,
+      "loss": 0.004,
+      "step": 74620
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.30925925925926e-05,
+      "loss": 0.0047,
+      "step": 74630
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.3091666666666666e-05,
+      "loss": 0.0019,
+      "step": 74640
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.309074074074074e-05,
+      "loss": 0.0025,
+      "step": 74650
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.3089814814814816e-05,
+      "loss": 0.0037,
+      "step": 74660
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.308888888888889e-05,
+      "loss": 0.0024,
+      "step": 74670
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.3087962962962966e-05,
+      "loss": 0.0062,
+      "step": 74680
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.3087037037037035e-05,
+      "loss": 0.0026,
+      "step": 74690
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.308611111111111e-05,
+      "loss": 0.0028,
+      "step": 74700
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.308518518518519e-05,
+      "loss": 0.0027,
+      "step": 74710
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.3084259259259266e-05,
+      "loss": 0.0034,
+      "step": 74720
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.3083333333333335e-05,
+      "loss": 0.0036,
+      "step": 74730
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.308240740740741e-05,
+      "loss": 0.0021,
+      "step": 74740
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.3081481481481485e-05,
+      "loss": 0.0051,
+      "step": 74750
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.308055555555556e-05,
+      "loss": 0.0024,
+      "step": 74760
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.307962962962963e-05,
+      "loss": 0.0032,
+      "step": 74770
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.30787037037037e-05,
+      "loss": 0.0046,
+      "step": 74780
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.307777777777778e-05,
+      "loss": 0.0075,
+      "step": 74790
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.307685185185185e-05,
+      "loss": 0.003,
+      "step": 74800
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.307592592592593e-05,
+      "loss": 0.0029,
+      "step": 74810
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.3075000000000003e-05,
+      "loss": 0.0024,
+      "step": 74820
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.307407407407408e-05,
+      "loss": 0.0041,
+      "step": 74830
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.3073148148148154e-05,
+      "loss": 0.0035,
+      "step": 74840
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.307222222222222e-05,
+      "loss": 0.0027,
+      "step": 74850
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.30712962962963e-05,
+      "loss": 0.0018,
+      "step": 74860
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.307037037037037e-05,
+      "loss": 0.0036,
+      "step": 74870
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.306944444444445e-05,
+      "loss": 0.0044,
+      "step": 74880
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.306851851851852e-05,
+      "loss": 0.0052,
+      "step": 74890
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.306759259259259e-05,
+      "loss": 0.002,
+      "step": 74900
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.3066666666666665e-05,
+      "loss": 0.0029,
+      "step": 74910
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.306574074074075e-05,
+      "loss": 0.004,
+      "step": 74920
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.306481481481482e-05,
+      "loss": 0.0031,
+      "step": 74930
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.306388888888889e-05,
+      "loss": 0.0036,
+      "step": 74940
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.3062962962962966e-05,
+      "loss": 0.003,
+      "step": 74950
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.306203703703704e-05,
+      "loss": 0.0032,
+      "step": 74960
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.3061111111111116e-05,
+      "loss": 0.0042,
+      "step": 74970
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.3060185185185184e-05,
+      "loss": 0.0019,
+      "step": 74980
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.305925925925926e-05,
+      "loss": 0.002,
+      "step": 74990
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.3058333333333334e-05,
+      "loss": 0.0023,
+      "step": 75000
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.305740740740741e-05,
+      "loss": 0.0016,
+      "step": 75010
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.3056481481481484e-05,
+      "loss": 0.003,
+      "step": 75020
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.305555555555556e-05,
+      "loss": 0.0027,
+      "step": 75030
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.3054629629629634e-05,
+      "loss": 0.0034,
+      "step": 75040
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.305370370370371e-05,
+      "loss": 0.0032,
+      "step": 75050
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.305277777777778e-05,
+      "loss": 0.0027,
+      "step": 75060
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.305185185185185e-05,
+      "loss": 0.0034,
+      "step": 75070
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.305092592592593e-05,
+      "loss": 0.0024,
+      "step": 75080
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.305e-05,
+      "loss": 0.0035,
+      "step": 75090
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.304907407407408e-05,
+      "loss": 0.0026,
+      "step": 75100
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.3048148148148146e-05,
+      "loss": 0.003,
+      "step": 75110
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.304722222222222e-05,
+      "loss": 0.0029,
+      "step": 75120
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.3046296296296296e-05,
+      "loss": 0.0017,
+      "step": 75130
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.304537037037038e-05,
+      "loss": 0.0017,
+      "step": 75140
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.3044444444444446e-05,
+      "loss": 0.0019,
+      "step": 75150
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.304351851851852e-05,
+      "loss": 0.0034,
+      "step": 75160
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.3042592592592597e-05,
+      "loss": 0.0057,
+      "step": 75170
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.304166666666667e-05,
+      "loss": 0.0026,
+      "step": 75180
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.304074074074074e-05,
+      "loss": 0.002,
+      "step": 75190
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.3039814814814815e-05,
+      "loss": 0.0037,
+      "step": 75200
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.303888888888889e-05,
+      "loss": 0.0044,
+      "step": 75210
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.3037962962962965e-05,
+      "loss": 0.0026,
+      "step": 75220
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.303703703703704e-05,
+      "loss": 0.0028,
+      "step": 75230
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.3036111111111115e-05,
+      "loss": 0.0021,
+      "step": 75240
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.303518518518519e-05,
+      "loss": 0.0029,
+      "step": 75250
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.3034351851851855e-05,
+      "loss": 0.0037,
+      "step": 75260
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.303342592592593e-05,
+      "loss": 0.0021,
+      "step": 75270
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.3032500000000005e-05,
+      "loss": 0.0042,
+      "step": 75280
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.303157407407408e-05,
+      "loss": 0.0016,
+      "step": 75290
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.303064814814815e-05,
+      "loss": 0.0014,
+      "step": 75300
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.3029722222222224e-05,
+      "loss": 0.0069,
+      "step": 75310
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.30287962962963e-05,
+      "loss": 0.0036,
+      "step": 75320
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.3027870370370374e-05,
+      "loss": 0.0038,
+      "step": 75330
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.302694444444444e-05,
+      "loss": 0.0047,
+      "step": 75340
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.302601851851852e-05,
+      "loss": 0.0022,
+      "step": 75350
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.30250925925926e-05,
+      "loss": 0.003,
+      "step": 75360
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.3024166666666674e-05,
+      "loss": 0.0034,
+      "step": 75370
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.302324074074074e-05,
+      "loss": 0.0027,
+      "step": 75380
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.302231481481482e-05,
+      "loss": 0.0026,
+      "step": 75390
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.302138888888889e-05,
+      "loss": 0.0018,
+      "step": 75400
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.302046296296297e-05,
+      "loss": 0.0026,
+      "step": 75410
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.3019537037037036e-05,
+      "loss": 0.0033,
+      "step": 75420
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.301861111111111e-05,
+      "loss": 0.0023,
+      "step": 75430
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.3017685185185186e-05,
+      "loss": 0.0018,
+      "step": 75440
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.301675925925926e-05,
+      "loss": 0.0012,
+      "step": 75450
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.3015833333333336e-05,
+      "loss": 0.0022,
+      "step": 75460
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.301490740740741e-05,
+      "loss": 0.0064,
+      "step": 75470
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.3013981481481486e-05,
+      "loss": 0.005,
+      "step": 75480
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.301305555555556e-05,
+      "loss": 0.0021,
+      "step": 75490
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.3012129629629636e-05,
+      "loss": 0.0023,
+      "step": 75500
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.3011203703703704e-05,
+      "loss": 0.0021,
+      "step": 75510
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.301027777777778e-05,
+      "loss": 0.0036,
+      "step": 75520
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.3009351851851854e-05,
+      "loss": 0.0022,
+      "step": 75530
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.300842592592593e-05,
+      "loss": 0.0032,
+      "step": 75540
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.30075e-05,
+      "loss": 0.0044,
+      "step": 75550
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.300657407407407e-05,
+      "loss": 0.005,
+      "step": 75560
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.300564814814815e-05,
+      "loss": 0.0027,
+      "step": 75570
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.300472222222223e-05,
+      "loss": 0.0023,
+      "step": 75580
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.30037962962963e-05,
+      "loss": 0.001,
+      "step": 75590
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.300287037037037e-05,
+      "loss": 0.0019,
+      "step": 75600
+    },
+    {
+      "epoch": 14.0,
+      "eval_loss": 0.014557932503521442,
+      "eval_runtime": 153.5155,
+      "eval_samples_per_second": 25.991,
+      "eval_steps_per_second": 3.25,
+      "step": 75600
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.300194444444445e-05,
+      "loss": 0.0015,
+      "step": 75610
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.300101851851852e-05,
+      "loss": 0.0034,
+      "step": 75620
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 4.300009259259259e-05,
+      "loss": 0.0018,
+      "step": 75630
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 4.2999166666666666e-05,
+      "loss": 0.0037,
+      "step": 75640
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 4.299824074074074e-05,
+      "loss": 0.0029,
+      "step": 75650
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 4.2997314814814817e-05,
+      "loss": 0.0018,
+      "step": 75660
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 4.299638888888889e-05,
+      "loss": 0.0017,
+      "step": 75670
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 4.299546296296296e-05,
+      "loss": 0.0051,
+      "step": 75680
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 4.299453703703704e-05,
+      "loss": 0.004,
+      "step": 75690
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 4.299361111111112e-05,
+      "loss": 0.0014,
+      "step": 75700
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 4.299268518518519e-05,
+      "loss": 0.0008,
+      "step": 75710
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 4.299175925925926e-05,
+      "loss": 0.0041,
+      "step": 75720
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 4.2990833333333335e-05,
+      "loss": 0.0032,
+      "step": 75730
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 4.298990740740741e-05,
+      "loss": 0.0048,
+      "step": 75740
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 4.2988981481481485e-05,
+      "loss": 0.0022,
+      "step": 75750
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 4.2988055555555554e-05,
+      "loss": 0.0028,
+      "step": 75760
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 4.298712962962963e-05,
+      "loss": 0.0015,
+      "step": 75770
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 4.2986203703703704e-05,
+      "loss": 0.0032,
+      "step": 75780
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 4.298527777777778e-05,
+      "loss": 0.0032,
+      "step": 75790
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 4.2984351851851854e-05,
+      "loss": 0.0015,
+      "step": 75800
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 4.298342592592593e-05,
+      "loss": 0.0032,
+      "step": 75810
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 4.2982500000000004e-05,
+      "loss": 0.0023,
+      "step": 75820
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 4.298157407407408e-05,
+      "loss": 0.0061,
+      "step": 75830
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 4.298064814814815e-05,
+      "loss": 0.0028,
+      "step": 75840
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 4.297972222222222e-05,
+      "loss": 0.0014,
+      "step": 75850
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 4.29787962962963e-05,
+      "loss": 0.0047,
+      "step": 75860
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 4.297787037037037e-05,
+      "loss": 0.0015,
+      "step": 75870
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 4.297694444444445e-05,
+      "loss": 0.0045,
+      "step": 75880
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 4.2976018518518516e-05,
+      "loss": 0.0021,
+      "step": 75890
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 4.29750925925926e-05,
+      "loss": 0.0047,
+      "step": 75900
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 4.297416666666667e-05,
+      "loss": 0.0018,
+      "step": 75910
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 4.297324074074075e-05,
+      "loss": 0.0025,
+      "step": 75920
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 4.2972314814814816e-05,
+      "loss": 0.0026,
+      "step": 75930
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 4.297138888888889e-05,
+      "loss": 0.0025,
+      "step": 75940
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 4.2970462962962966e-05,
+      "loss": 0.005,
+      "step": 75950
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 4.296953703703704e-05,
+      "loss": 0.0023,
+      "step": 75960
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 4.296861111111111e-05,
+      "loss": 0.0029,
+      "step": 75970
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 4.2967685185185184e-05,
+      "loss": 0.0036,
+      "step": 75980
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 4.296675925925926e-05,
+      "loss": 0.0021,
+      "step": 75990
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 4.2965833333333335e-05,
+      "loss": 0.0018,
+      "step": 76000
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 4.296490740740741e-05,
+      "loss": 0.0031,
+      "step": 76010
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 4.2963981481481485e-05,
+      "loss": 0.0024,
+      "step": 76020
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 4.296305555555556e-05,
+      "loss": 0.0032,
+      "step": 76030
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 4.2962129629629635e-05,
+      "loss": 0.0043,
+      "step": 76040
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 4.29612037037037e-05,
+      "loss": 0.0028,
+      "step": 76050
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 4.296027777777778e-05,
+      "loss": 0.0043,
+      "step": 76060
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 4.295935185185185e-05,
+      "loss": 0.0029,
+      "step": 76070
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 4.295842592592593e-05,
+      "loss": 0.0029,
+      "step": 76080
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 4.29575e-05,
+      "loss": 0.0012,
+      "step": 76090
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 4.295657407407407e-05,
+      "loss": 0.0031,
+      "step": 76100
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 4.295564814814815e-05,
+      "loss": 0.0033,
+      "step": 76110
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 4.295472222222223e-05,
+      "loss": 0.0024,
+      "step": 76120
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 4.2953796296296304e-05,
+      "loss": 0.0024,
+      "step": 76130
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 4.295287037037037e-05,
+      "loss": 0.0045,
+      "step": 76140
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 4.295194444444445e-05,
+      "loss": 0.0025,
+      "step": 76150
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 4.295101851851852e-05,
+      "loss": 0.0026,
+      "step": 76160
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 4.29500925925926e-05,
+      "loss": 0.0028,
+      "step": 76170
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 4.2949166666666665e-05,
+      "loss": 0.0034,
+      "step": 76180
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 4.294824074074074e-05,
+      "loss": 0.002,
+      "step": 76190
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 4.2947314814814815e-05,
+      "loss": 0.0039,
+      "step": 76200
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 4.294638888888889e-05,
+      "loss": 0.0035,
+      "step": 76210
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 4.2945462962962965e-05,
+      "loss": 0.0026,
+      "step": 76220
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 4.294453703703704e-05,
+      "loss": 0.0033,
+      "step": 76230
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 4.2943611111111116e-05,
+      "loss": 0.0051,
+      "step": 76240
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 4.294268518518519e-05,
+      "loss": 0.0026,
+      "step": 76250
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 4.294175925925926e-05,
+      "loss": 0.0027,
+      "step": 76260
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 4.2940833333333334e-05,
+      "loss": 0.0031,
+      "step": 76270
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 4.293990740740741e-05,
+      "loss": 0.0047,
+      "step": 76280
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 4.2938981481481484e-05,
+      "loss": 0.0022,
+      "step": 76290
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 4.293805555555556e-05,
+      "loss": 0.0062,
+      "step": 76300
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 4.293712962962963e-05,
+      "loss": 0.0022,
+      "step": 76310
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 4.29362037037037e-05,
+      "loss": 0.0019,
+      "step": 76320
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 4.2935277777777784e-05,
+      "loss": 0.0019,
+      "step": 76330
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 4.293435185185186e-05,
+      "loss": 0.001,
+      "step": 76340
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 4.293342592592593e-05,
+      "loss": 0.0015,
+      "step": 76350
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 4.29325e-05,
+      "loss": 0.0026,
+      "step": 76360
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 4.293157407407408e-05,
+      "loss": 0.0015,
+      "step": 76370
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 4.293064814814815e-05,
+      "loss": 0.002,
+      "step": 76380
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 4.292972222222222e-05,
+      "loss": 0.0032,
+      "step": 76390
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 4.2928796296296296e-05,
+      "loss": 0.0045,
+      "step": 76400
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 4.292787037037037e-05,
+      "loss": 0.0021,
+      "step": 76410
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 4.2926944444444446e-05,
+      "loss": 0.0021,
+      "step": 76420
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 4.292601851851852e-05,
+      "loss": 0.0031,
+      "step": 76430
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 4.2925092592592596e-05,
+      "loss": 0.0078,
+      "step": 76440
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 4.292416666666667e-05,
+      "loss": 0.0023,
+      "step": 76450
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 4.2923240740740747e-05,
+      "loss": 0.0034,
+      "step": 76460
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 4.2922314814814815e-05,
+      "loss": 0.0022,
+      "step": 76470
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 4.292138888888889e-05,
+      "loss": 0.0064,
+      "step": 76480
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 4.2920462962962965e-05,
+      "loss": 0.0031,
+      "step": 76490
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 4.291953703703704e-05,
+      "loss": 0.0024,
+      "step": 76500
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 4.2918611111111115e-05,
+      "loss": 0.0027,
+      "step": 76510
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 4.291768518518518e-05,
+      "loss": 0.0049,
+      "step": 76520
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 4.291675925925926e-05,
+      "loss": 0.0014,
+      "step": 76530
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 4.2915833333333333e-05,
+      "loss": 0.0026,
+      "step": 76540
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 4.2914907407407415e-05,
+      "loss": 0.0044,
+      "step": 76550
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 4.2913981481481484e-05,
+      "loss": 0.0032,
+      "step": 76560
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 4.291305555555556e-05,
+      "loss": 0.0021,
+      "step": 76570
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 4.2912129629629634e-05,
+      "loss": 0.003,
+      "step": 76580
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 4.291120370370371e-05,
+      "loss": 0.0016,
+      "step": 76590
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 4.291027777777778e-05,
+      "loss": 0.0019,
+      "step": 76600
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 4.290935185185185e-05,
+      "loss": 0.0031,
+      "step": 76610
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 4.290842592592593e-05,
+      "loss": 0.003,
+      "step": 76620
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 4.29075e-05,
+      "loss": 0.0017,
+      "step": 76630
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 4.290657407407408e-05,
+      "loss": 0.0035,
+      "step": 76640
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 4.2905648148148145e-05,
+      "loss": 0.002,
+      "step": 76650
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 4.290472222222223e-05,
+      "loss": 0.002,
+      "step": 76660
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 4.29037962962963e-05,
+      "loss": 0.0024,
+      "step": 76670
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 4.290287037037037e-05,
+      "loss": 0.0039,
+      "step": 76680
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 4.2901944444444446e-05,
+      "loss": 0.0028,
+      "step": 76690
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 4.290101851851852e-05,
+      "loss": 0.002,
+      "step": 76700
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 4.2900092592592596e-05,
+      "loss": 0.004,
+      "step": 76710
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 4.289916666666667e-05,
+      "loss": 0.0047,
+      "step": 76720
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 4.289824074074074e-05,
+      "loss": 0.003,
+      "step": 76730
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 4.2897314814814814e-05,
+      "loss": 0.0017,
+      "step": 76740
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 4.289638888888889e-05,
+      "loss": 0.0023,
+      "step": 76750
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 4.2895462962962964e-05,
+      "loss": 0.002,
+      "step": 76760
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 4.289453703703704e-05,
+      "loss": 0.0059,
+      "step": 76770
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 4.2893611111111114e-05,
+      "loss": 0.0025,
+      "step": 76780
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 4.289268518518519e-05,
+      "loss": 0.0027,
+      "step": 76790
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 4.2891759259259265e-05,
+      "loss": 0.0027,
+      "step": 76800
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 4.289083333333333e-05,
+      "loss": 0.0025,
+      "step": 76810
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 4.288990740740741e-05,
+      "loss": 0.0036,
+      "step": 76820
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 4.288898148148148e-05,
+      "loss": 0.0024,
+      "step": 76830
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 4.288805555555556e-05,
+      "loss": 0.0029,
+      "step": 76840
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 4.288712962962963e-05,
+      "loss": 0.0036,
+      "step": 76850
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 4.28862037037037e-05,
+      "loss": 0.0026,
+      "step": 76860
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 4.288527777777778e-05,
+      "loss": 0.0022,
+      "step": 76870
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 4.288435185185186e-05,
+      "loss": 0.0021,
+      "step": 76880
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 4.2883425925925926e-05,
+      "loss": 0.0025,
+      "step": 76890
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 4.28825e-05,
+      "loss": 0.0017,
+      "step": 76900
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 4.2881574074074077e-05,
+      "loss": 0.0009,
+      "step": 76910
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 4.288064814814815e-05,
+      "loss": 0.0026,
+      "step": 76920
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 4.287972222222223e-05,
+      "loss": 0.0037,
+      "step": 76930
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 4.2878796296296295e-05,
+      "loss": 0.0021,
+      "step": 76940
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 4.287787037037037e-05,
+      "loss": 0.003,
+      "step": 76950
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 4.2876944444444445e-05,
+      "loss": 0.0029,
+      "step": 76960
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 4.287601851851852e-05,
+      "loss": 0.0049,
+      "step": 76970
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 4.2875092592592595e-05,
+      "loss": 0.0026,
+      "step": 76980
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 4.287416666666667e-05,
+      "loss": 0.0059,
+      "step": 76990
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 4.2873240740740745e-05,
+      "loss": 0.0012,
+      "step": 77000
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 4.287231481481482e-05,
+      "loss": 0.0021,
+      "step": 77010
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 4.287138888888889e-05,
+      "loss": 0.0032,
+      "step": 77020
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 4.2870462962962964e-05,
+      "loss": 0.0013,
+      "step": 77030
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 4.286953703703704e-05,
+      "loss": 0.0045,
+      "step": 77040
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 4.2868611111111114e-05,
+      "loss": 0.0037,
+      "step": 77050
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 4.286768518518519e-05,
+      "loss": 0.0019,
+      "step": 77060
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 4.286675925925926e-05,
+      "loss": 0.0026,
+      "step": 77070
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 4.286583333333333e-05,
+      "loss": 0.0014,
+      "step": 77080
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 4.2864907407407414e-05,
+      "loss": 0.0027,
+      "step": 77090
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 4.286398148148148e-05,
+      "loss": 0.0013,
+      "step": 77100
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 4.286305555555556e-05,
+      "loss": 0.0032,
+      "step": 77110
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 4.286212962962963e-05,
+      "loss": 0.0026,
+      "step": 77120
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 4.286120370370371e-05,
+      "loss": 0.0012,
+      "step": 77130
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 4.286027777777778e-05,
+      "loss": 0.0035,
+      "step": 77140
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 4.285935185185185e-05,
+      "loss": 0.0043,
+      "step": 77150
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 4.2858425925925926e-05,
+      "loss": 0.0024,
+      "step": 77160
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 4.28575e-05,
+      "loss": 0.0032,
+      "step": 77170
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 4.2856574074074076e-05,
+      "loss": 0.0017,
+      "step": 77180
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 4.2855648148148144e-05,
+      "loss": 0.0027,
+      "step": 77190
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 4.2854722222222226e-05,
+      "loss": 0.0014,
+      "step": 77200
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 4.28537962962963e-05,
+      "loss": 0.0039,
+      "step": 77210
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 4.2852870370370376e-05,
+      "loss": 0.0033,
+      "step": 77220
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 4.2851944444444444e-05,
+      "loss": 0.0032,
+      "step": 77230
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 4.285101851851852e-05,
+      "loss": 0.0034,
+      "step": 77240
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 4.2850092592592595e-05,
+      "loss": 0.0017,
+      "step": 77250
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 4.284916666666667e-05,
+      "loss": 0.002,
+      "step": 77260
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 4.2848240740740745e-05,
+      "loss": 0.0028,
+      "step": 77270
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 4.284731481481481e-05,
+      "loss": 0.0012,
+      "step": 77280
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 4.284638888888889e-05,
+      "loss": 0.0026,
+      "step": 77290
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 4.284546296296296e-05,
+      "loss": 0.0029,
+      "step": 77300
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 4.284453703703704e-05,
+      "loss": 0.0027,
+      "step": 77310
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 4.284361111111111e-05,
+      "loss": 0.0034,
+      "step": 77320
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 4.284268518518519e-05,
+      "loss": 0.0011,
+      "step": 77330
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 4.284175925925926e-05,
+      "loss": 0.0023,
+      "step": 77340
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 4.284083333333334e-05,
+      "loss": 0.0014,
+      "step": 77350
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 4.283990740740741e-05,
+      "loss": 0.0028,
+      "step": 77360
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 4.283898148148148e-05,
+      "loss": 0.0026,
+      "step": 77370
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 4.283805555555556e-05,
+      "loss": 0.0042,
+      "step": 77380
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 4.283712962962963e-05,
+      "loss": 0.0023,
+      "step": 77390
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 4.28362037037037e-05,
+      "loss": 0.0045,
+      "step": 77400
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 4.283527777777778e-05,
+      "loss": 0.0023,
+      "step": 77410
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 4.283435185185186e-05,
+      "loss": 0.0041,
+      "step": 77420
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 4.283342592592593e-05,
+      "loss": 0.0022,
+      "step": 77430
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 4.28325e-05,
+      "loss": 0.0038,
+      "step": 77440
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 4.2831574074074075e-05,
+      "loss": 0.0024,
+      "step": 77450
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 4.283064814814815e-05,
+      "loss": 0.0031,
+      "step": 77460
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 4.2829722222222225e-05,
+      "loss": 0.0009,
+      "step": 77470
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 4.28287962962963e-05,
+      "loss": 0.0028,
+      "step": 77480
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 4.282787037037037e-05,
+      "loss": 0.0031,
+      "step": 77490
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 4.2826944444444444e-05,
+      "loss": 0.0024,
+      "step": 77500
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 4.282601851851852e-05,
+      "loss": 0.0017,
+      "step": 77510
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 4.2825092592592594e-05,
+      "loss": 0.0039,
+      "step": 77520
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 4.282416666666667e-05,
+      "loss": 0.0015,
+      "step": 77530
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 4.2823240740740744e-05,
+      "loss": 0.0024,
+      "step": 77540
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 4.282231481481482e-05,
+      "loss": 0.0026,
+      "step": 77550
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 4.2821388888888894e-05,
+      "loss": 0.004,
+      "step": 77560
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 4.282046296296296e-05,
+      "loss": 0.0034,
+      "step": 77570
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 4.281953703703704e-05,
+      "loss": 0.0035,
+      "step": 77580
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 4.281861111111111e-05,
+      "loss": 0.0043,
+      "step": 77590
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 4.281768518518519e-05,
+      "loss": 0.0022,
+      "step": 77600
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 4.2816759259259256e-05,
+      "loss": 0.0029,
+      "step": 77610
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 4.281583333333333e-05,
+      "loss": 0.0023,
+      "step": 77620
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 4.281490740740741e-05,
+      "loss": 0.0044,
+      "step": 77630
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 4.281398148148149e-05,
+      "loss": 0.0037,
+      "step": 77640
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 4.2813055555555556e-05,
+      "loss": 0.0028,
+      "step": 77650
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 4.281212962962963e-05,
+      "loss": 0.0017,
+      "step": 77660
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 4.2811203703703706e-05,
+      "loss": 0.0061,
+      "step": 77670
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 4.281027777777778e-05,
+      "loss": 0.0022,
+      "step": 77680
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 4.2809351851851856e-05,
+      "loss": 0.0032,
+      "step": 77690
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 4.2808425925925925e-05,
+      "loss": 0.0017,
+      "step": 77700
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 4.28075e-05,
+      "loss": 0.002,
+      "step": 77710
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 4.2806574074074075e-05,
+      "loss": 0.0026,
+      "step": 77720
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 4.280564814814815e-05,
+      "loss": 0.0024,
+      "step": 77730
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 4.2804722222222225e-05,
+      "loss": 0.0039,
+      "step": 77740
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 4.28037962962963e-05,
+      "loss": 0.0039,
+      "step": 77750
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 4.2802962962962965e-05,
+      "loss": 0.0035,
+      "step": 77760
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 4.280203703703704e-05,
+      "loss": 0.0061,
+      "step": 77770
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 4.2801111111111115e-05,
+      "loss": 0.0033,
+      "step": 77780
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 4.280018518518519e-05,
+      "loss": 0.0019,
+      "step": 77790
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 4.279925925925926e-05,
+      "loss": 0.0024,
+      "step": 77800
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 4.279833333333333e-05,
+      "loss": 0.0049,
+      "step": 77810
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 4.279740740740741e-05,
+      "loss": 0.0019,
+      "step": 77820
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 4.279648148148148e-05,
+      "loss": 0.0057,
+      "step": 77830
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 4.279555555555556e-05,
+      "loss": 0.0022,
+      "step": 77840
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 4.279462962962963e-05,
+      "loss": 0.0029,
+      "step": 77850
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 4.279370370370371e-05,
+      "loss": 0.0041,
+      "step": 77860
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 4.2792777777777784e-05,
+      "loss": 0.0041,
+      "step": 77870
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 4.279185185185185e-05,
+      "loss": 0.002,
+      "step": 77880
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 4.279092592592593e-05,
+      "loss": 0.0028,
+      "step": 77890
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 4.279e-05,
+      "loss": 0.0025,
+      "step": 77900
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 4.278907407407408e-05,
+      "loss": 0.0023,
+      "step": 77910
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 4.278814814814815e-05,
+      "loss": 0.0029,
+      "step": 77920
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 4.278722222222222e-05,
+      "loss": 0.0045,
+      "step": 77930
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 4.2786296296296295e-05,
+      "loss": 0.0028,
+      "step": 77940
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 4.278537037037037e-05,
+      "loss": 0.0021,
+      "step": 77950
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 4.278444444444445e-05,
+      "loss": 0.002,
+      "step": 77960
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 4.278351851851852e-05,
+      "loss": 0.0017,
+      "step": 77970
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 4.2782592592592596e-05,
+      "loss": 0.0026,
+      "step": 77980
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 4.278166666666667e-05,
+      "loss": 0.0031,
+      "step": 77990
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 4.2780740740740746e-05,
+      "loss": 0.0026,
+      "step": 78000
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 4.2779814814814814e-05,
+      "loss": 0.0022,
+      "step": 78010
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 4.277888888888889e-05,
+      "loss": 0.0022,
+      "step": 78020
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 4.2777962962962964e-05,
+      "loss": 0.0034,
+      "step": 78030
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 4.277703703703704e-05,
+      "loss": 0.0009,
+      "step": 78040
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 4.2776111111111114e-05,
+      "loss": 0.0019,
+      "step": 78050
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 4.277518518518518e-05,
+      "loss": 0.003,
+      "step": 78060
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 4.2774259259259264e-05,
+      "loss": 0.0016,
+      "step": 78070
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 4.277333333333334e-05,
+      "loss": 0.0028,
+      "step": 78080
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 4.277240740740741e-05,
+      "loss": 0.0026,
+      "step": 78090
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 4.277148148148148e-05,
+      "loss": 0.0042,
+      "step": 78100
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 4.277055555555556e-05,
+      "loss": 0.0043,
+      "step": 78110
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 4.276962962962963e-05,
+      "loss": 0.0012,
+      "step": 78120
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 4.276870370370371e-05,
+      "loss": 0.0033,
+      "step": 78130
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 4.2767777777777776e-05,
+      "loss": 0.0008,
+      "step": 78140
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 4.276685185185185e-05,
+      "loss": 0.0034,
+      "step": 78150
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 4.2765925925925926e-05,
+      "loss": 0.0035,
+      "step": 78160
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 4.2765e-05,
+      "loss": 0.0038,
+      "step": 78170
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 4.2764074074074076e-05,
+      "loss": 0.0043,
+      "step": 78180
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 4.276314814814815e-05,
+      "loss": 0.0057,
+      "step": 78190
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 4.2762222222222227e-05,
+      "loss": 0.0028,
+      "step": 78200
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 4.27612962962963e-05,
+      "loss": 0.0036,
+      "step": 78210
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 4.276037037037037e-05,
+      "loss": 0.0035,
+      "step": 78220
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 4.2759444444444445e-05,
+      "loss": 0.0042,
+      "step": 78230
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 4.275851851851852e-05,
+      "loss": 0.0031,
+      "step": 78240
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 4.2757592592592595e-05,
+      "loss": 0.0032,
+      "step": 78250
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 4.275666666666667e-05,
+      "loss": 0.0042,
+      "step": 78260
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 4.275574074074074e-05,
+      "loss": 0.0027,
+      "step": 78270
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 4.2754814814814813e-05,
+      "loss": 0.0027,
+      "step": 78280
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 4.2753888888888895e-05,
+      "loss": 0.0044,
+      "step": 78290
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 4.2752962962962964e-05,
+      "loss": 0.0041,
+      "step": 78300
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 4.275203703703704e-05,
+      "loss": 0.0024,
+      "step": 78310
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 4.2751111111111114e-05,
+      "loss": 0.0059,
+      "step": 78320
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 4.275018518518519e-05,
+      "loss": 0.0029,
+      "step": 78330
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 4.2749259259259264e-05,
+      "loss": 0.0043,
+      "step": 78340
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 4.274833333333333e-05,
+      "loss": 0.0022,
+      "step": 78350
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 4.274740740740741e-05,
+      "loss": 0.0036,
+      "step": 78360
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 4.274648148148148e-05,
+      "loss": 0.0044,
+      "step": 78370
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 4.274555555555556e-05,
+      "loss": 0.0027,
+      "step": 78380
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 4.274462962962963e-05,
+      "loss": 0.0019,
+      "step": 78390
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 4.274370370370371e-05,
+      "loss": 0.0034,
+      "step": 78400
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 4.274277777777778e-05,
+      "loss": 0.0024,
+      "step": 78410
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 4.274185185185186e-05,
+      "loss": 0.0028,
+      "step": 78420
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 4.2740925925925926e-05,
+      "loss": 0.0016,
+      "step": 78430
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 4.274e-05,
+      "loss": 0.0023,
+      "step": 78440
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 4.2739074074074076e-05,
+      "loss": 0.0071,
+      "step": 78450
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 4.273814814814815e-05,
+      "loss": 0.0023,
+      "step": 78460
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 4.2737222222222226e-05,
+      "loss": 0.0023,
+      "step": 78470
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 4.2736296296296294e-05,
+      "loss": 0.0052,
+      "step": 78480
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 4.273537037037037e-05,
+      "loss": 0.0031,
+      "step": 78490
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 4.273444444444445e-05,
+      "loss": 0.0025,
+      "step": 78500
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 4.273351851851852e-05,
+      "loss": 0.004,
+      "step": 78510
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 4.2732592592592594e-05,
+      "loss": 0.0024,
+      "step": 78520
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 4.273166666666667e-05,
+      "loss": 0.0029,
+      "step": 78530
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 4.2730740740740745e-05,
+      "loss": 0.0024,
+      "step": 78540
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 4.272981481481482e-05,
+      "loss": 0.0016,
+      "step": 78550
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 4.272888888888889e-05,
+      "loss": 0.004,
+      "step": 78560
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 4.272796296296296e-05,
+      "loss": 0.0017,
+      "step": 78570
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 4.272703703703704e-05,
+      "loss": 0.0027,
+      "step": 78580
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 4.272611111111111e-05,
+      "loss": 0.0042,
+      "step": 78590
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 4.272518518518519e-05,
+      "loss": 0.0017,
+      "step": 78600
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 4.272425925925926e-05,
+      "loss": 0.0024,
+      "step": 78610
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 4.272333333333334e-05,
+      "loss": 0.0017,
+      "step": 78620
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 4.272240740740741e-05,
+      "loss": 0.0018,
+      "step": 78630
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 4.272148148148148e-05,
+      "loss": 0.0026,
+      "step": 78640
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 4.2720555555555557e-05,
+      "loss": 0.003,
+      "step": 78650
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 4.271962962962963e-05,
+      "loss": 0.0053,
+      "step": 78660
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 4.271870370370371e-05,
+      "loss": 0.0033,
+      "step": 78670
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 4.271777777777778e-05,
+      "loss": 0.0017,
+      "step": 78680
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 4.271685185185185e-05,
+      "loss": 0.0026,
+      "step": 78690
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 4.2715925925925925e-05,
+      "loss": 0.004,
+      "step": 78700
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 4.2715e-05,
+      "loss": 0.0036,
+      "step": 78710
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 4.2714074074074075e-05,
+      "loss": 0.004,
+      "step": 78720
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 4.271314814814815e-05,
+      "loss": 0.0032,
+      "step": 78730
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 4.2712222222222225e-05,
+      "loss": 0.0022,
+      "step": 78740
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 4.27112962962963e-05,
+      "loss": 0.0023,
+      "step": 78750
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 4.2710370370370375e-05,
+      "loss": 0.0034,
+      "step": 78760
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 4.2709444444444444e-05,
+      "loss": 0.0037,
+      "step": 78770
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 4.270851851851852e-05,
+      "loss": 0.0033,
+      "step": 78780
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 4.2707592592592594e-05,
+      "loss": 0.0025,
+      "step": 78790
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 4.270666666666667e-05,
+      "loss": 0.0023,
+      "step": 78800
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 4.2705740740740744e-05,
+      "loss": 0.0031,
+      "step": 78810
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 4.270481481481481e-05,
+      "loss": 0.0022,
+      "step": 78820
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 4.2703888888888894e-05,
+      "loss": 0.0013,
+      "step": 78830
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 4.270296296296297e-05,
+      "loss": 0.003,
+      "step": 78840
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 4.270203703703704e-05,
+      "loss": 0.0033,
+      "step": 78850
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 4.270111111111111e-05,
+      "loss": 0.0018,
+      "step": 78860
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 4.270018518518519e-05,
+      "loss": 0.0014,
+      "step": 78870
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 4.269925925925926e-05,
+      "loss": 0.0016,
+      "step": 78880
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 4.269833333333334e-05,
+      "loss": 0.0058,
+      "step": 78890
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 4.2697407407407406e-05,
+      "loss": 0.0027,
+      "step": 78900
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 4.269648148148148e-05,
+      "loss": 0.004,
+      "step": 78910
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 4.2695555555555556e-05,
+      "loss": 0.0019,
+      "step": 78920
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 4.269462962962963e-05,
+      "loss": 0.0028,
+      "step": 78930
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 4.2693703703703706e-05,
+      "loss": 0.003,
+      "step": 78940
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 4.269277777777778e-05,
+      "loss": 0.0041,
+      "step": 78950
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 4.2691851851851856e-05,
+      "loss": 0.0025,
+      "step": 78960
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 4.269092592592593e-05,
+      "loss": 0.0026,
+      "step": 78970
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 4.269e-05,
+      "loss": 0.0049,
+      "step": 78980
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 4.2689074074074075e-05,
+      "loss": 0.0025,
+      "step": 78990
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 4.268814814814815e-05,
+      "loss": 0.0038,
+      "step": 79000
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 4.2687222222222225e-05,
+      "loss": 0.0022,
+      "step": 79010
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 4.26862962962963e-05,
+      "loss": 0.005,
+      "step": 79020
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 4.268537037037037e-05,
+      "loss": 0.0041,
+      "step": 79030
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 4.268444444444445e-05,
+      "loss": 0.0018,
+      "step": 79040
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 4.2683518518518525e-05,
+      "loss": 0.0033,
+      "step": 79050
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 4.268259259259259e-05,
+      "loss": 0.0021,
+      "step": 79060
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 4.268166666666667e-05,
+      "loss": 0.0012,
+      "step": 79070
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 4.268074074074074e-05,
+      "loss": 0.0043,
+      "step": 79080
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 4.267981481481482e-05,
+      "loss": 0.0037,
+      "step": 79090
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 4.2678888888888893e-05,
+      "loss": 0.0015,
+      "step": 79100
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 4.267796296296296e-05,
+      "loss": 0.0034,
+      "step": 79110
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 4.267703703703704e-05,
+      "loss": 0.0021,
+      "step": 79120
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 4.267611111111111e-05,
+      "loss": 0.0025,
+      "step": 79130
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 4.267518518518519e-05,
+      "loss": 0.0021,
+      "step": 79140
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 4.267425925925926e-05,
+      "loss": 0.0017,
+      "step": 79150
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 4.267333333333334e-05,
+      "loss": 0.002,
+      "step": 79160
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 4.267240740740741e-05,
+      "loss": 0.0031,
+      "step": 79170
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 4.267148148148149e-05,
+      "loss": 0.0014,
+      "step": 79180
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 4.2670555555555555e-05,
+      "loss": 0.0047,
+      "step": 79190
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 4.266962962962963e-05,
+      "loss": 0.0018,
+      "step": 79200
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 4.2668703703703705e-05,
+      "loss": 0.0036,
+      "step": 79210
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 4.266777777777778e-05,
+      "loss": 0.0017,
+      "step": 79220
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 4.2666851851851856e-05,
+      "loss": 0.0024,
+      "step": 79230
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 4.2665925925925924e-05,
+      "loss": 0.0024,
+      "step": 79240
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 4.2665e-05,
+      "loss": 0.007,
+      "step": 79250
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 4.266407407407408e-05,
+      "loss": 0.0022,
+      "step": 79260
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 4.266314814814815e-05,
+      "loss": 0.0023,
+      "step": 79270
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 4.2662222222222224e-05,
+      "loss": 0.0018,
+      "step": 79280
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 4.26612962962963e-05,
+      "loss": 0.003,
+      "step": 79290
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 4.2660370370370374e-05,
+      "loss": 0.0035,
+      "step": 79300
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 4.265944444444445e-05,
+      "loss": 0.0039,
+      "step": 79310
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 4.265851851851852e-05,
+      "loss": 0.0037,
+      "step": 79320
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 4.265759259259259e-05,
+      "loss": 0.0028,
+      "step": 79330
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 4.265666666666667e-05,
+      "loss": 0.0027,
+      "step": 79340
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 4.265574074074074e-05,
+      "loss": 0.0033,
+      "step": 79350
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 4.265481481481481e-05,
+      "loss": 0.0017,
+      "step": 79360
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 4.265388888888889e-05,
+      "loss": 0.0019,
+      "step": 79370
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 4.265296296296297e-05,
+      "loss": 0.0046,
+      "step": 79380
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 4.265203703703704e-05,
+      "loss": 0.003,
+      "step": 79390
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 4.265111111111111e-05,
+      "loss": 0.0045,
+      "step": 79400
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 4.2650185185185186e-05,
+      "loss": 0.0035,
+      "step": 79410
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 4.264925925925926e-05,
+      "loss": 0.0011,
+      "step": 79420
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 4.2648333333333336e-05,
+      "loss": 0.0033,
+      "step": 79430
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 4.264740740740741e-05,
+      "loss": 0.0043,
+      "step": 79440
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 4.264648148148148e-05,
+      "loss": 0.0042,
+      "step": 79450
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 4.2645555555555555e-05,
+      "loss": 0.0029,
+      "step": 79460
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 4.2644629629629637e-05,
+      "loss": 0.0032,
+      "step": 79470
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 4.2643703703703705e-05,
+      "loss": 0.002,
+      "step": 79480
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 4.264277777777778e-05,
+      "loss": 0.0026,
+      "step": 79490
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 4.2641851851851855e-05,
+      "loss": 0.0036,
+      "step": 79500
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 4.264092592592593e-05,
+      "loss": 0.0017,
+      "step": 79510
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 4.2640000000000005e-05,
+      "loss": 0.0019,
+      "step": 79520
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 4.263907407407407e-05,
+      "loss": 0.0017,
+      "step": 79530
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 4.263814814814815e-05,
+      "loss": 0.004,
+      "step": 79540
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 4.2637222222222223e-05,
+      "loss": 0.0025,
+      "step": 79550
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 4.26362962962963e-05,
+      "loss": 0.0011,
+      "step": 79560
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 4.263537037037037e-05,
+      "loss": 0.0022,
+      "step": 79570
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 4.263444444444445e-05,
+      "loss": 0.003,
+      "step": 79580
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 4.2633518518518524e-05,
+      "loss": 0.0014,
+      "step": 79590
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 4.26325925925926e-05,
+      "loss": 0.002,
+      "step": 79600
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 4.263166666666667e-05,
+      "loss": 0.0024,
+      "step": 79610
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 4.263074074074074e-05,
+      "loss": 0.0022,
+      "step": 79620
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 4.262981481481482e-05,
+      "loss": 0.0057,
+      "step": 79630
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 4.262888888888889e-05,
+      "loss": 0.0042,
+      "step": 79640
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 4.262796296296297e-05,
+      "loss": 0.0024,
+      "step": 79650
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 4.2627037037037036e-05,
+      "loss": 0.0046,
+      "step": 79660
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 4.262611111111111e-05,
+      "loss": 0.0022,
+      "step": 79670
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 4.2625185185185186e-05,
+      "loss": 0.0035,
+      "step": 79680
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 4.262425925925926e-05,
+      "loss": 0.0033,
+      "step": 79690
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 4.2623333333333336e-05,
+      "loss": 0.004,
+      "step": 79700
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 4.262240740740741e-05,
+      "loss": 0.0016,
+      "step": 79710
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 4.2621481481481486e-05,
+      "loss": 0.0034,
+      "step": 79720
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 4.262055555555556e-05,
+      "loss": 0.0033,
+      "step": 79730
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 4.261962962962963e-05,
+      "loss": 0.0026,
+      "step": 79740
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 4.2618703703703704e-05,
+      "loss": 0.0036,
+      "step": 79750
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 4.261777777777778e-05,
+      "loss": 0.0015,
+      "step": 79760
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 4.2616851851851854e-05,
+      "loss": 0.0026,
+      "step": 79770
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 4.261592592592592e-05,
+      "loss": 0.0021,
+      "step": 79780
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 4.2615e-05,
+      "loss": 0.0037,
+      "step": 79790
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 4.261407407407408e-05,
+      "loss": 0.0021,
+      "step": 79800
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 4.2613148148148155e-05,
+      "loss": 0.0036,
+      "step": 79810
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 4.261222222222222e-05,
+      "loss": 0.0025,
+      "step": 79820
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 4.26112962962963e-05,
+      "loss": 0.0026,
+      "step": 79830
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 4.261037037037037e-05,
+      "loss": 0.0031,
+      "step": 79840
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 4.260944444444445e-05,
+      "loss": 0.0009,
+      "step": 79850
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 4.260851851851852e-05,
+      "loss": 0.0024,
+      "step": 79860
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 4.260759259259259e-05,
+      "loss": 0.0018,
+      "step": 79870
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 4.2606666666666666e-05,
+      "loss": 0.0048,
+      "step": 79880
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 4.260574074074074e-05,
+      "loss": 0.0041,
+      "step": 79890
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 4.2604814814814817e-05,
+      "loss": 0.0027,
+      "step": 79900
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 4.260388888888889e-05,
+      "loss": 0.003,
+      "step": 79910
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 4.260296296296297e-05,
+      "loss": 0.0033,
+      "step": 79920
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 4.260203703703704e-05,
+      "loss": 0.0036,
+      "step": 79930
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 4.260111111111112e-05,
+      "loss": 0.0047,
+      "step": 79940
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 4.2600185185185185e-05,
+      "loss": 0.0022,
+      "step": 79950
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 4.259925925925926e-05,
+      "loss": 0.0029,
+      "step": 79960
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 4.2598333333333335e-05,
+      "loss": 0.0021,
+      "step": 79970
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 4.259740740740741e-05,
+      "loss": 0.0032,
+      "step": 79980
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 4.259648148148148e-05,
+      "loss": 0.0015,
+      "step": 79990
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 4.2595555555555554e-05,
+      "loss": 0.0021,
+      "step": 80000
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 4.2594629629629635e-05,
+      "loss": 0.0041,
+      "step": 80010
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 4.259370370370371e-05,
+      "loss": 0.004,
+      "step": 80020
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 4.259277777777778e-05,
+      "loss": 0.0022,
+      "step": 80030
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 4.2591851851851854e-05,
+      "loss": 0.0023,
+      "step": 80040
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 4.259092592592593e-05,
+      "loss": 0.0038,
+      "step": 80050
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 4.2590000000000004e-05,
+      "loss": 0.0019,
+      "step": 80060
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 4.258907407407408e-05,
+      "loss": 0.0017,
+      "step": 80070
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 4.258814814814815e-05,
+      "loss": 0.0018,
+      "step": 80080
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 4.258722222222222e-05,
+      "loss": 0.0033,
+      "step": 80090
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 4.25862962962963e-05,
+      "loss": 0.0034,
+      "step": 80100
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 4.258537037037037e-05,
+      "loss": 0.0009,
+      "step": 80110
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 4.258444444444445e-05,
+      "loss": 0.0046,
+      "step": 80120
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 4.258351851851852e-05,
+      "loss": 0.0056,
+      "step": 80130
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 4.25825925925926e-05,
+      "loss": 0.0033,
+      "step": 80140
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 4.258166666666667e-05,
+      "loss": 0.0037,
+      "step": 80150
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 4.258074074074074e-05,
+      "loss": 0.0019,
+      "step": 80160
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 4.2579814814814816e-05,
+      "loss": 0.0036,
+      "step": 80170
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 4.257888888888889e-05,
+      "loss": 0.0036,
+      "step": 80180
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 4.2577962962962966e-05,
+      "loss": 0.0017,
+      "step": 80190
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 4.2577037037037034e-05,
+      "loss": 0.0015,
+      "step": 80200
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 4.257611111111111e-05,
+      "loss": 0.0025,
+      "step": 80210
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 4.2575185185185184e-05,
+      "loss": 0.0031,
+      "step": 80220
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 4.2574259259259266e-05,
+      "loss": 0.0037,
+      "step": 80230
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 4.2573333333333335e-05,
+      "loss": 0.0024,
+      "step": 80240
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 4.257240740740741e-05,
+      "loss": 0.0045,
+      "step": 80250
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 4.2571481481481485e-05,
+      "loss": 0.0037,
+      "step": 80260
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 4.257055555555556e-05,
+      "loss": 0.003,
+      "step": 80270
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 4.2569629629629635e-05,
+      "loss": 0.0024,
+      "step": 80280
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 4.25687037037037e-05,
+      "loss": 0.0031,
+      "step": 80290
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 4.256777777777778e-05,
+      "loss": 0.0018,
+      "step": 80300
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 4.256685185185185e-05,
+      "loss": 0.0013,
+      "step": 80310
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 4.256592592592593e-05,
+      "loss": 0.0029,
+      "step": 80320
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 4.2564999999999997e-05,
+      "loss": 0.0013,
+      "step": 80330
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 4.256407407407408e-05,
+      "loss": 0.0032,
+      "step": 80340
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 4.256324074074074e-05,
+      "loss": 0.0023,
+      "step": 80350
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 4.256231481481482e-05,
+      "loss": 0.0026,
+      "step": 80360
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 4.256138888888889e-05,
+      "loss": 0.0023,
+      "step": 80370
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 4.256046296296297e-05,
+      "loss": 0.0031,
+      "step": 80380
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 4.255953703703704e-05,
+      "loss": 0.0016,
+      "step": 80390
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 4.255861111111111e-05,
+      "loss": 0.0031,
+      "step": 80400
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 4.255768518518519e-05,
+      "loss": 0.0047,
+      "step": 80410
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 4.255675925925926e-05,
+      "loss": 0.0022,
+      "step": 80420
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 4.255583333333334e-05,
+      "loss": 0.0021,
+      "step": 80430
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 4.2554907407407405e-05,
+      "loss": 0.0025,
+      "step": 80440
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 4.255398148148148e-05,
+      "loss": 0.0016,
+      "step": 80450
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 4.255305555555556e-05,
+      "loss": 0.0022,
+      "step": 80460
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 4.255212962962963e-05,
+      "loss": 0.0013,
+      "step": 80470
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 4.2551203703703705e-05,
+      "loss": 0.0043,
+      "step": 80480
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 4.255027777777778e-05,
+      "loss": 0.003,
+      "step": 80490
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 4.2549351851851855e-05,
+      "loss": 0.0022,
+      "step": 80500
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 4.254842592592593e-05,
+      "loss": 0.0038,
+      "step": 80510
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 4.25475e-05,
+      "loss": 0.0032,
+      "step": 80520
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 4.2546574074074074e-05,
+      "loss": 0.0053,
+      "step": 80530
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 4.254564814814815e-05,
+      "loss": 0.0022,
+      "step": 80540
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 4.2544722222222224e-05,
+      "loss": 0.0027,
+      "step": 80550
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 4.25437962962963e-05,
+      "loss": 0.0022,
+      "step": 80560
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 4.2542870370370374e-05,
+      "loss": 0.0019,
+      "step": 80570
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 4.254194444444445e-05,
+      "loss": 0.0023,
+      "step": 80580
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 4.2541018518518524e-05,
+      "loss": 0.0026,
+      "step": 80590
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 4.254009259259259e-05,
+      "loss": 0.0046,
+      "step": 80600
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 4.253916666666667e-05,
+      "loss": 0.0025,
+      "step": 80610
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 4.253824074074074e-05,
+      "loss": 0.0027,
+      "step": 80620
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 4.253731481481482e-05,
+      "loss": 0.005,
+      "step": 80630
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 4.253638888888889e-05,
+      "loss": 0.0026,
+      "step": 80640
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 4.253546296296296e-05,
+      "loss": 0.0034,
+      "step": 80650
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 4.2534537037037036e-05,
+      "loss": 0.0029,
+      "step": 80660
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 4.253361111111112e-05,
+      "loss": 0.0016,
+      "step": 80670
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 4.2532685185185186e-05,
+      "loss": 0.0046,
+      "step": 80680
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 4.253175925925926e-05,
+      "loss": 0.003,
+      "step": 80690
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 4.2530833333333336e-05,
+      "loss": 0.0042,
+      "step": 80700
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 4.252990740740741e-05,
+      "loss": 0.0034,
+      "step": 80710
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 4.2528981481481486e-05,
+      "loss": 0.0028,
+      "step": 80720
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 4.2528055555555555e-05,
+      "loss": 0.0019,
+      "step": 80730
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 4.252712962962963e-05,
+      "loss": 0.0015,
+      "step": 80740
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 4.2526203703703705e-05,
+      "loss": 0.0052,
+      "step": 80750
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 4.252527777777778e-05,
+      "loss": 0.0034,
+      "step": 80760
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 4.252435185185185e-05,
+      "loss": 0.0031,
+      "step": 80770
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 4.252342592592593e-05,
+      "loss": 0.0032,
+      "step": 80780
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 4.2522500000000005e-05,
+      "loss": 0.0018,
+      "step": 80790
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 4.252157407407408e-05,
+      "loss": 0.0029,
+      "step": 80800
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 4.252064814814815e-05,
+      "loss": 0.004,
+      "step": 80810
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 4.251972222222222e-05,
+      "loss": 0.0023,
+      "step": 80820
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 4.25187962962963e-05,
+      "loss": 0.0049,
+      "step": 80830
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 4.2517870370370373e-05,
+      "loss": 0.0025,
+      "step": 80840
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 4.251694444444445e-05,
+      "loss": 0.0031,
+      "step": 80850
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 4.251601851851852e-05,
+      "loss": 0.0047,
+      "step": 80860
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 4.251509259259259e-05,
+      "loss": 0.0036,
+      "step": 80870
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 4.251416666666667e-05,
+      "loss": 0.0025,
+      "step": 80880
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 4.251324074074074e-05,
+      "loss": 0.0043,
+      "step": 80890
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 4.251231481481482e-05,
+      "loss": 0.0011,
+      "step": 80900
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 4.251138888888889e-05,
+      "loss": 0.0025,
+      "step": 80910
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 4.251046296296297e-05,
+      "loss": 0.0017,
+      "step": 80920
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 4.250962962962963e-05,
+      "loss": 0.0025,
+      "step": 80930
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 4.250870370370371e-05,
+      "loss": 0.0035,
+      "step": 80940
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 4.250777777777778e-05,
+      "loss": 0.0037,
+      "step": 80950
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 4.250685185185185e-05,
+      "loss": 0.0033,
+      "step": 80960
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 4.2505925925925925e-05,
+      "loss": 0.0027,
+      "step": 80970
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 4.2505e-05,
+      "loss": 0.0012,
+      "step": 80980
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 4.2504074074074076e-05,
+      "loss": 0.0045,
+      "step": 80990
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 4.250314814814815e-05,
+      "loss": 0.0024,
+      "step": 81000
+    },
+    {
+      "epoch": 15.0,
+      "eval_loss": 0.015327811241149902,
+      "eval_runtime": 153.5155,
+      "eval_samples_per_second": 25.991,
+      "eval_steps_per_second": 3.25,
+      "step": 81000
+    },
+    {
+      "epoch": 15.0,
+      "step": 81000,
+      "total_flos": 7.546037374716461e+19,
+      "train_loss": 0.0059461380455321,
+      "train_runtime": 98084.884,
+      "train_samples_per_second": 44.043,
+      "train_steps_per_second": 5.505
+    }
+  ],
+  "max_steps": 540000,
+  "num_train_epochs": 100,
+  "total_flos": 7.546037374716461e+19,
+  "trial_name": null,
+  "trial_params": null
+}