diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,18998 @@
+{
+  "best_metric": 1.1188170909881592,
+  "best_model_checkpoint": "/work/twskvfb446/facebook/wav2vec2-large-lv60_voidful/phoneme_byt5_SpeechMixEEDT5_w2v2-large_t5lephone-small_lrdiff2_bs64/checkpoint-31521",
+  "epoch": 7.0,
+  "global_step": 31521,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.714285714285715e-07,
+      "loss": 4.8338,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.142857142857143e-06,
+      "loss": 4.8403,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7142857142857145e-06,
+      "loss": 4.6805,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.285714285714286e-06,
+      "loss": 4.5098,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 4.3628,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.428571428571429e-06,
+      "loss": 4.3564,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 4.1557,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.571428571428572e-06,
+      "loss": 4.0656,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.142857142857142e-06,
+      "loss": 3.8855,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 3.7873,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.285714285714286e-06,
+      "loss": 3.685,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.857142857142858e-06,
+      "loss": 3.5988,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.428571428571429e-06,
+      "loss": 3.4437,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 3.3101,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 3.2639,
+      "step": 150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.142857142857144e-06,
+      "loss": 3.1035,
+      "step": 160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.714285714285715e-06,
+      "loss": 2.9991,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0285714285714285e-05,
+      "loss": 2.9539,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0857142857142858e-05,
+      "loss": 2.941,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 2.8688,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2e-05,
+      "loss": 2.8255,
+      "step": 210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2571428571428572e-05,
+      "loss": 2.7809,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3142857142857145e-05,
+      "loss": 2.7848,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3714285714285716e-05,
+      "loss": 2.6878,
+      "step": 240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4285714285714287e-05,
+      "loss": 2.6746,
+      "step": 250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4857142857142858e-05,
+      "loss": 2.6664,
+      "step": 260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.542857142857143e-05,
+      "loss": 2.5992,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 2.5776,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.6571428571428574e-05,
+      "loss": 2.5659,
+      "step": 290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 2.5339,
+      "step": 300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7714285714285717e-05,
+      "loss": 2.5111,
+      "step": 310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8285714285714288e-05,
+      "loss": 2.5012,
+      "step": 320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.885714285714286e-05,
+      "loss": 2.4786,
+      "step": 330
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.942857142857143e-05,
+      "loss": 2.4416,
+      "step": 340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2e-05,
+      "loss": 2.4435,
+      "step": 350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.057142857142857e-05,
+      "loss": 2.4357,
+      "step": 360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1142857142857144e-05,
+      "loss": 2.397,
+      "step": 370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1714285714285715e-05,
+      "loss": 2.401,
+      "step": 380
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.2285714285714287e-05,
+      "loss": 2.375,
+      "step": 390
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.2857142857142858e-05,
+      "loss": 2.3784,
+      "step": 400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.3428571428571433e-05,
+      "loss": 2.3549,
+      "step": 410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.4e-05,
+      "loss": 2.3334,
+      "step": 420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.4571428571428575e-05,
+      "loss": 2.3282,
+      "step": 430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5142857142857143e-05,
+      "loss": 2.3087,
+      "step": 440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.5714285714285718e-05,
+      "loss": 2.3273,
+      "step": 450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.628571428571429e-05,
+      "loss": 2.2403,
+      "step": 460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.6857142857142857e-05,
+      "loss": 2.1895,
+      "step": 470
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.742857142857143e-05,
+      "loss": 2.1575,
+      "step": 480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8e-05,
+      "loss": 2.1236,
+      "step": 490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8571428571428574e-05,
+      "loss": 2.0825,
+      "step": 500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9142857142857146e-05,
+      "loss": 2.0988,
+      "step": 510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9714285714285717e-05,
+      "loss": 2.0474,
+      "step": 520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.0285714285714288e-05,
+      "loss": 2.0416,
+      "step": 530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.085714285714286e-05,
+      "loss": 2.002,
+      "step": 540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.142857142857143e-05,
+      "loss": 2.0199,
+      "step": 550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 1.9793,
+      "step": 560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.257142857142857e-05,
+      "loss": 1.9666,
+      "step": 570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.314285714285715e-05,
+      "loss": 1.9722,
+      "step": 580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.3714285714285716e-05,
+      "loss": 1.9398,
+      "step": 590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.4285714285714284e-05,
+      "loss": 1.9595,
+      "step": 600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.485714285714286e-05,
+      "loss": 1.97,
+      "step": 610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.542857142857143e-05,
+      "loss": 1.9175,
+      "step": 620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.6e-05,
+      "loss": 1.9196,
+      "step": 630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.6571428571428576e-05,
+      "loss": 1.9297,
+      "step": 640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.714285714285715e-05,
+      "loss": 1.8824,
+      "step": 650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.771428571428572e-05,
+      "loss": 1.9058,
+      "step": 660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.828571428571429e-05,
+      "loss": 1.8983,
+      "step": 670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.885714285714286e-05,
+      "loss": 1.8565,
+      "step": 680
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.9428571428571435e-05,
+      "loss": 1.8626,
+      "step": 690
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4e-05,
+      "loss": 1.8494,
+      "step": 700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.999702358806459e-05,
+      "loss": 1.8598,
+      "step": 710
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.999404717612918e-05,
+      "loss": 1.8434,
+      "step": 720
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.999107076419377e-05,
+      "loss": 1.8494,
+      "step": 730
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.998809435225836e-05,
+      "loss": 1.7983,
+      "step": 740
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.9985117940322945e-05,
+      "loss": 1.8372,
+      "step": 750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.998214152838753e-05,
+      "loss": 1.8138,
+      "step": 760
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.997916511645212e-05,
+      "loss": 1.8005,
+      "step": 770
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.9976188704516705e-05,
+      "loss": 1.8083,
+      "step": 780
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.99732122925813e-05,
+      "loss": 1.8047,
+      "step": 790
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.9970235880645886e-05,
+      "loss": 1.8119,
+      "step": 800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.996725946871047e-05,
+      "loss": 1.7993,
+      "step": 810
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.996428305677506e-05,
+      "loss": 1.8174,
+      "step": 820
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.996130664483965e-05,
+      "loss": 1.7907,
+      "step": 830
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.995833023290424e-05,
+      "loss": 1.7951,
+      "step": 840
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.995535382096883e-05,
+      "loss": 1.7833,
+      "step": 850
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.9952377409033414e-05,
+      "loss": 1.7893,
+      "step": 860
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.9949400997098e-05,
+      "loss": 1.7739,
+      "step": 870
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.994642458516259e-05,
+      "loss": 1.7959,
+      "step": 880
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9943448173227175e-05,
+      "loss": 1.7773,
+      "step": 890
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.994047176129177e-05,
+      "loss": 1.769,
+      "step": 900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9937495349356355e-05,
+      "loss": 1.7596,
+      "step": 910
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.993451893742094e-05,
+      "loss": 1.7659,
+      "step": 920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.993154252548553e-05,
+      "loss": 1.7592,
+      "step": 930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.992856611355012e-05,
+      "loss": 1.7695,
+      "step": 940
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.992558970161471e-05,
+      "loss": 1.7539,
+      "step": 950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.992261328967929e-05,
+      "loss": 1.7511,
+      "step": 960
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9919636877743883e-05,
+      "loss": 1.7402,
+      "step": 970
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.991666046580847e-05,
+      "loss": 1.7562,
+      "step": 980
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.991368405387306e-05,
+      "loss": 1.7448,
+      "step": 990
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9910707641937644e-05,
+      "loss": 1.751,
+      "step": 1000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.990773123000224e-05,
+      "loss": 1.735,
+      "step": 1010
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.9904754818066825e-05,
+      "loss": 1.7277,
+      "step": 1020
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.990177840613141e-05,
+      "loss": 1.743,
+      "step": 1030
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.9898801994196e-05,
+      "loss": 1.7349,
+      "step": 1040
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.989582558226059e-05,
+      "loss": 1.735,
+      "step": 1050
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.989284917032517e-05,
+      "loss": 1.7349,
+      "step": 1060
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.9889872758389766e-05,
+      "loss": 1.7154,
+      "step": 1070
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.988689634645435e-05,
+      "loss": 1.7371,
+      "step": 1080
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.988391993451894e-05,
+      "loss": 1.7301,
+      "step": 1090
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.988094352258353e-05,
+      "loss": 1.7114,
+      "step": 1100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.987796711064812e-05,
+      "loss": 1.7207,
+      "step": 1110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.987499069871271e-05,
+      "loss": 1.7162,
+      "step": 1120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.9872014286777294e-05,
+      "loss": 1.715,
+      "step": 1130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.986903787484188e-05,
+      "loss": 1.728,
+      "step": 1140
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.986606146290647e-05,
+      "loss": 1.7043,
+      "step": 1150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.986308505097106e-05,
+      "loss": 1.6983,
+      "step": 1160
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.986010863903564e-05,
+      "loss": 1.6977,
+      "step": 1170
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.9857132227100235e-05,
+      "loss": 1.7026,
+      "step": 1180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.985415581516482e-05,
+      "loss": 1.719,
+      "step": 1190
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.985117940322941e-05,
+      "loss": 1.6955,
+      "step": 1200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.9848202991293996e-05,
+      "loss": 1.6891,
+      "step": 1210
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.984522657935859e-05,
+      "loss": 1.7063,
+      "step": 1220
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.984225016742318e-05,
+      "loss": 1.6837,
+      "step": 1230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.9839273755487764e-05,
+      "loss": 1.7061,
+      "step": 1240
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.983629734355235e-05,
+      "loss": 1.693,
+      "step": 1250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.9833320931616944e-05,
+      "loss": 1.6938,
+      "step": 1260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.9830344519681524e-05,
+      "loss": 1.6636,
+      "step": 1270
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.982736810774611e-05,
+      "loss": 1.6865,
+      "step": 1280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.9824391695810705e-05,
+      "loss": 1.682,
+      "step": 1290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.982141528387529e-05,
+      "loss": 1.6979,
+      "step": 1300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.981843887193988e-05,
+      "loss": 1.6958,
+      "step": 1310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.9815462460004466e-05,
+      "loss": 1.6674,
+      "step": 1320
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.981248604806906e-05,
+      "loss": 1.6978,
+      "step": 1330
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9809509636133646e-05,
+      "loss": 1.6704,
+      "step": 1340
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.980653322419823e-05,
+      "loss": 1.6825,
+      "step": 1350
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.980355681226282e-05,
+      "loss": 1.6565,
+      "step": 1360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.980058040032741e-05,
+      "loss": 1.6619,
+      "step": 1370
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9797603988391994e-05,
+      "loss": 1.6971,
+      "step": 1380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.979462757645658e-05,
+      "loss": 1.6852,
+      "step": 1390
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9791651164521174e-05,
+      "loss": 1.6619,
+      "step": 1400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.978867475258576e-05,
+      "loss": 1.6702,
+      "step": 1410
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.978569834065035e-05,
+      "loss": 1.6727,
+      "step": 1420
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9782721928714935e-05,
+      "loss": 1.6758,
+      "step": 1430
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.977974551677953e-05,
+      "loss": 1.6611,
+      "step": 1440
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9776769104844116e-05,
+      "loss": 1.6655,
+      "step": 1450
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.97737926929087e-05,
+      "loss": 1.6648,
+      "step": 1460
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.977081628097329e-05,
+      "loss": 1.6597,
+      "step": 1470
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9767839869037876e-05,
+      "loss": 1.6605,
+      "step": 1480
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.976486345710246e-05,
+      "loss": 1.6496,
+      "step": 1490
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.976188704516706e-05,
+      "loss": 1.6576,
+      "step": 1500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.9758910633231644e-05,
+      "loss": 1.6491,
+      "step": 1510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.975593422129623e-05,
+      "loss": 1.6556,
+      "step": 1520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.975295780936082e-05,
+      "loss": 1.6661,
+      "step": 1530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.974998139742541e-05,
+      "loss": 1.6495,
+      "step": 1540
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.974700498549e-05,
+      "loss": 1.6522,
+      "step": 1550
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.9744028573554585e-05,
+      "loss": 1.6375,
+      "step": 1560
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.974105216161917e-05,
+      "loss": 1.6383,
+      "step": 1570
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.973807574968376e-05,
+      "loss": 1.6532,
+      "step": 1580
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.9735099337748346e-05,
+      "loss": 1.6463,
+      "step": 1590
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.973212292581293e-05,
+      "loss": 1.649,
+      "step": 1600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9729146513877526e-05,
+      "loss": 1.6093,
+      "step": 1610
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.972617010194211e-05,
+      "loss": 1.6211,
+      "step": 1620
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.97231936900067e-05,
+      "loss": 1.6387,
+      "step": 1630
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.972021727807129e-05,
+      "loss": 1.6311,
+      "step": 1640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.971724086613588e-05,
+      "loss": 1.6282,
+      "step": 1650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.971426445420047e-05,
+      "loss": 1.6344,
+      "step": 1660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.9711288042265054e-05,
+      "loss": 1.6427,
+      "step": 1670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.970831163032964e-05,
+      "loss": 1.6177,
+      "step": 1680
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.970533521839423e-05,
+      "loss": 1.6341,
+      "step": 1690
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9702358806458815e-05,
+      "loss": 1.6331,
+      "step": 1700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.96993823945234e-05,
+      "loss": 1.6227,
+      "step": 1710
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9696405982587996e-05,
+      "loss": 1.6084,
+      "step": 1720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.969342957065258e-05,
+      "loss": 1.6176,
+      "step": 1730
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.969045315871717e-05,
+      "loss": 1.6214,
+      "step": 1740
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.9687476746781756e-05,
+      "loss": 1.6157,
+      "step": 1750
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.968450033484635e-05,
+      "loss": 1.6166,
+      "step": 1760
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.968152392291094e-05,
+      "loss": 1.6362,
+      "step": 1770
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.967854751097552e-05,
+      "loss": 1.617,
+      "step": 1780
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.967557109904011e-05,
+      "loss": 1.6175,
+      "step": 1790
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.96725946871047e-05,
+      "loss": 1.6246,
+      "step": 1800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.9669618275169285e-05,
+      "loss": 1.6406,
+      "step": 1810
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.966664186323387e-05,
+      "loss": 1.6243,
+      "step": 1820
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.9663665451298465e-05,
+      "loss": 1.6223,
+      "step": 1830
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.966068903936305e-05,
+      "loss": 1.6198,
+      "step": 1840
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.965771262742764e-05,
+      "loss": 1.6278,
+      "step": 1850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.9654736215492226e-05,
+      "loss": 1.6187,
+      "step": 1860
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.965175980355682e-05,
+      "loss": 1.6135,
+      "step": 1870
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.96487833916214e-05,
+      "loss": 1.6055,
+      "step": 1880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.964580697968599e-05,
+      "loss": 1.6063,
+      "step": 1890
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.964283056775058e-05,
+      "loss": 1.6158,
+      "step": 1900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.963985415581517e-05,
+      "loss": 1.6267,
+      "step": 1910
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.9636877743879754e-05,
+      "loss": 1.5936,
+      "step": 1920
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.963390133194435e-05,
+      "loss": 1.6001,
+      "step": 1930
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.9630924920008935e-05,
+      "loss": 1.5773,
+      "step": 1940
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.962794850807352e-05,
+      "loss": 1.6196,
+      "step": 1950
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.962497209613811e-05,
+      "loss": 1.5988,
+      "step": 1960
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.96219956842027e-05,
+      "loss": 1.606,
+      "step": 1970
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.961901927226728e-05,
+      "loss": 1.5979,
+      "step": 1980
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.961604286033187e-05,
+      "loss": 1.6236,
+      "step": 1990
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.961306644839646e-05,
+      "loss": 1.6158,
+      "step": 2000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.961009003646105e-05,
+      "loss": 1.6075,
+      "step": 2010
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.9607113624525637e-05,
+      "loss": 1.5907,
+      "step": 2020
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.9604137212590223e-05,
+      "loss": 1.5795,
+      "step": 2030
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.960116080065482e-05,
+      "loss": 1.5991,
+      "step": 2040
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.9598184388719404e-05,
+      "loss": 1.6036,
+      "step": 2050
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.959520797678399e-05,
+      "loss": 1.5972,
+      "step": 2060
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.959223156484858e-05,
+      "loss": 1.5929,
+      "step": 2070
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.958925515291317e-05,
+      "loss": 1.6135,
+      "step": 2080
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.958627874097775e-05,
+      "loss": 1.5965,
+      "step": 2090
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.958330232904234e-05,
+      "loss": 1.5721,
+      "step": 2100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.958032591710693e-05,
+      "loss": 1.6161,
+      "step": 2110
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.957734950517152e-05,
+      "loss": 1.5773,
+      "step": 2120
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.9574373093236106e-05,
+      "loss": 1.5842,
+      "step": 2130
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.957139668130069e-05,
+      "loss": 1.5768,
+      "step": 2140
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.9568420269365287e-05,
+      "loss": 1.5952,
+      "step": 2150
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.9565443857429873e-05,
+      "loss": 1.5914,
+      "step": 2160
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.956246744549446e-05,
+      "loss": 1.5858,
+      "step": 2170
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.955949103355905e-05,
+      "loss": 1.5981,
+      "step": 2180
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.9556514621623634e-05,
+      "loss": 1.5979,
+      "step": 2190
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.955353820968822e-05,
+      "loss": 1.5776,
+      "step": 2200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.955056179775281e-05,
+      "loss": 1.5858,
+      "step": 2210
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.95475853858174e-05,
+      "loss": 1.5765,
+      "step": 2220
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.954460897388199e-05,
+      "loss": 1.5679,
+      "step": 2230
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.9541632561946575e-05,
+      "loss": 1.5826,
+      "step": 2240
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.953865615001116e-05,
+      "loss": 1.5584,
+      "step": 2250
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.9535679738075756e-05,
+      "loss": 1.5761,
+      "step": 2260
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.953270332614034e-05,
+      "loss": 1.5726,
+      "step": 2270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.952972691420493e-05,
+      "loss": 1.5794,
+      "step": 2280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.952675050226952e-05,
+      "loss": 1.5713,
+      "step": 2290
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.9523774090334104e-05,
+      "loss": 1.5826,
+      "step": 2300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.952079767839869e-05,
+      "loss": 1.5817,
+      "step": 2310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.9517821266463284e-05,
+      "loss": 1.5893,
+      "step": 2320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.951484485452787e-05,
+      "loss": 1.5868,
+      "step": 2330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.951186844259246e-05,
+      "loss": 1.5877,
+      "step": 2340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.9508892030657045e-05,
+      "loss": 1.5833,
+      "step": 2350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.950591561872164e-05,
+      "loss": 1.5799,
+      "step": 2360
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.9502939206786225e-05,
+      "loss": 1.5572,
+      "step": 2370
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.949996279485081e-05,
+      "loss": 1.556,
+      "step": 2380
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.94969863829154e-05,
+      "loss": 1.5751,
+      "step": 2390
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.9494009970979986e-05,
+      "loss": 1.5972,
+      "step": 2400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.949103355904457e-05,
+      "loss": 1.5636,
+      "step": 2410
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.948805714710916e-05,
+      "loss": 1.5675,
+      "step": 2420
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.9485080735173754e-05,
+      "loss": 1.5717,
+      "step": 2430
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.948210432323834e-05,
+      "loss": 1.5882,
+      "step": 2440
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.947912791130293e-05,
+      "loss": 1.57,
+      "step": 2450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.9476151499367514e-05,
+      "loss": 1.5653,
+      "step": 2460
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.947317508743211e-05,
+      "loss": 1.5688,
+      "step": 2470
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.9470198675496695e-05,
+      "loss": 1.5653,
+      "step": 2480
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.946722226356128e-05,
+      "loss": 1.5763,
+      "step": 2490
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.946424585162587e-05,
+      "loss": 1.5734,
+      "step": 2500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.9461269439690456e-05,
+      "loss": 1.5544,
+      "step": 2510
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.945829302775504e-05,
+      "loss": 1.5677,
+      "step": 2520
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.945531661581963e-05,
+      "loss": 1.566,
+      "step": 2530
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.945234020388422e-05,
+      "loss": 1.5546,
+      "step": 2540
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.944936379194881e-05,
+      "loss": 1.5657,
+      "step": 2550
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.94463873800134e-05,
+      "loss": 1.5427,
+      "step": 2560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9443410968077984e-05,
+      "loss": 1.5625,
+      "step": 2570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.944043455614258e-05,
+      "loss": 1.5843,
+      "step": 2580
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9437458144207164e-05,
+      "loss": 1.5627,
+      "step": 2590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9434481732271744e-05,
+      "loss": 1.5557,
+      "step": 2600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.943150532033634e-05,
+      "loss": 1.5608,
+      "step": 2610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9428528908400925e-05,
+      "loss": 1.5534,
+      "step": 2620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.942555249646551e-05,
+      "loss": 1.5554,
+      "step": 2630
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.94225760845301e-05,
+      "loss": 1.5498,
+      "step": 2640
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.941959967259469e-05,
+      "loss": 1.5577,
+      "step": 2650
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.941662326065928e-05,
+      "loss": 1.5533,
+      "step": 2660
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9413646848723866e-05,
+      "loss": 1.5644,
+      "step": 2670
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.941067043678845e-05,
+      "loss": 1.5648,
+      "step": 2680
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.940769402485305e-05,
+      "loss": 1.5518,
+      "step": 2690
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.940471761291763e-05,
+      "loss": 1.5476,
+      "step": 2700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.940174120098222e-05,
+      "loss": 1.5732,
+      "step": 2710
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.939876478904681e-05,
+      "loss": 1.5329,
+      "step": 2720
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9395788377111394e-05,
+      "loss": 1.5428,
+      "step": 2730
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.939281196517598e-05,
+      "loss": 1.56,
+      "step": 2740
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9389835553240575e-05,
+      "loss": 1.5553,
+      "step": 2750
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.938685914130516e-05,
+      "loss": 1.55,
+      "step": 2760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.938388272936975e-05,
+      "loss": 1.5454,
+      "step": 2770
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9380906317434336e-05,
+      "loss": 1.5549,
+      "step": 2780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.937792990549893e-05,
+      "loss": 1.5397,
+      "step": 2790
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.937495349356351e-05,
+      "loss": 1.5402,
+      "step": 2800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9371977081628096e-05,
+      "loss": 1.5646,
+      "step": 2810
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.936900066969269e-05,
+      "loss": 1.5459,
+      "step": 2820
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.936602425775728e-05,
+      "loss": 1.539,
+      "step": 2830
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9363047845821864e-05,
+      "loss": 1.5546,
+      "step": 2840
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.936007143388645e-05,
+      "loss": 1.5612,
+      "step": 2850
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9357095021951044e-05,
+      "loss": 1.5497,
+      "step": 2860
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.935411861001563e-05,
+      "loss": 1.5549,
+      "step": 2870
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.935114219808022e-05,
+      "loss": 1.553,
+      "step": 2880
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9348165786144805e-05,
+      "loss": 1.5347,
+      "step": 2890
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.93451893742094e-05,
+      "loss": 1.5542,
+      "step": 2900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.934221296227398e-05,
+      "loss": 1.5547,
+      "step": 2910
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9339236550338566e-05,
+      "loss": 1.5431,
+      "step": 2920
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.933626013840316e-05,
+      "loss": 1.5507,
+      "step": 2930
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9333283726467746e-05,
+      "loss": 1.5369,
+      "step": 2940
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.933030731453233e-05,
+      "loss": 1.5259,
+      "step": 2950
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.932733090259692e-05,
+      "loss": 1.5419,
+      "step": 2960
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9324354490661514e-05,
+      "loss": 1.546,
+      "step": 2970
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.93213780787261e-05,
+      "loss": 1.5264,
+      "step": 2980
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.931840166679069e-05,
+      "loss": 1.5362,
+      "step": 2990
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.9315425254855275e-05,
+      "loss": 1.5343,
+      "step": 3000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.931244884291986e-05,
+      "loss": 1.534,
+      "step": 3010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.930947243098445e-05,
+      "loss": 1.5508,
+      "step": 3020
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.9306496019049035e-05,
+      "loss": 1.5422,
+      "step": 3030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.930351960711363e-05,
+      "loss": 1.5345,
+      "step": 3040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.9300543195178216e-05,
+      "loss": 1.5369,
+      "step": 3050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.92975667832428e-05,
+      "loss": 1.5227,
+      "step": 3060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.929459037130739e-05,
+      "loss": 1.5311,
+      "step": 3070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.929161395937198e-05,
+      "loss": 1.5557,
+      "step": 3080
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.928863754743657e-05,
+      "loss": 1.5382,
+      "step": 3090
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.928566113550116e-05,
+      "loss": 1.5438,
+      "step": 3100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.9282684723565744e-05,
+      "loss": 1.5174,
+      "step": 3110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.927970831163033e-05,
+      "loss": 1.5197,
+      "step": 3120
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.927673189969492e-05,
+      "loss": 1.5343,
+      "step": 3130
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.927375548775951e-05,
+      "loss": 1.5259,
+      "step": 3140
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.92707790758241e-05,
+      "loss": 1.5244,
+      "step": 3150
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.9267802663888685e-05,
+      "loss": 1.531,
+      "step": 3160
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.926482625195327e-05,
+      "loss": 1.5356,
+      "step": 3170
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9261849840017866e-05,
+      "loss": 1.5448,
+      "step": 3180
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.925887342808245e-05,
+      "loss": 1.5287,
+      "step": 3190
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.925589701614704e-05,
+      "loss": 1.5417,
+      "step": 3200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9252920604211627e-05,
+      "loss": 1.5189,
+      "step": 3210
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9249944192276213e-05,
+      "loss": 1.5303,
+      "step": 3220
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.92469677803408e-05,
+      "loss": 1.5242,
+      "step": 3230
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.924399136840539e-05,
+      "loss": 1.5226,
+      "step": 3240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.924101495646998e-05,
+      "loss": 1.5257,
+      "step": 3250
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.923803854453457e-05,
+      "loss": 1.5339,
+      "step": 3260
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.9235062132599155e-05,
+      "loss": 1.5305,
+      "step": 3270
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.923208572066374e-05,
+      "loss": 1.5434,
+      "step": 3280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.9229109308728335e-05,
+      "loss": 1.5342,
+      "step": 3290
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.922613289679292e-05,
+      "loss": 1.5278,
+      "step": 3300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.922315648485751e-05,
+      "loss": 1.5316,
+      "step": 3310
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.9220180072922096e-05,
+      "loss": 1.519,
+      "step": 3320
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.921720366098668e-05,
+      "loss": 1.5412,
+      "step": 3330
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.921422724905127e-05,
+      "loss": 1.5168,
+      "step": 3340
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.921125083711586e-05,
+      "loss": 1.5461,
+      "step": 3350
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.920827442518045e-05,
+      "loss": 1.523,
+      "step": 3360
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.920529801324504e-05,
+      "loss": 1.5262,
+      "step": 3370
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.9202321601309624e-05,
+      "loss": 1.5234,
+      "step": 3380
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.919934518937421e-05,
+      "loss": 1.5145,
+      "step": 3390
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.9196368777438805e-05,
+      "loss": 1.5092,
+      "step": 3400
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.919339236550339e-05,
+      "loss": 1.5192,
+      "step": 3410
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.919041595356797e-05,
+      "loss": 1.5141,
+      "step": 3420
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.9187439541632565e-05,
+      "loss": 1.5101,
+      "step": 3430
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.918446312969715e-05,
+      "loss": 1.5207,
+      "step": 3440
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.918148671776174e-05,
+      "loss": 1.518,
+      "step": 3450
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.9178510305826326e-05,
+      "loss": 1.5381,
+      "step": 3460
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.917553389389092e-05,
+      "loss": 1.5193,
+      "step": 3470
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.917255748195551e-05,
+      "loss": 1.5152,
+      "step": 3480
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9169581070020094e-05,
+      "loss": 1.5376,
+      "step": 3490
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.916660465808468e-05,
+      "loss": 1.5163,
+      "step": 3500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9163628246149274e-05,
+      "loss": 1.5132,
+      "step": 3510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9160651834213854e-05,
+      "loss": 1.5107,
+      "step": 3520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.915767542227845e-05,
+      "loss": 1.5345,
+      "step": 3530
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9154699010343035e-05,
+      "loss": 1.5292,
+      "step": 3540
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.915172259840762e-05,
+      "loss": 1.5181,
+      "step": 3550
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.914874618647221e-05,
+      "loss": 1.5143,
+      "step": 3560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.91457697745368e-05,
+      "loss": 1.4955,
+      "step": 3570
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.914279336260139e-05,
+      "loss": 1.5133,
+      "step": 3580
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9139816950665976e-05,
+      "loss": 1.5087,
+      "step": 3590
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.913684053873056e-05,
+      "loss": 1.5068,
+      "step": 3600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.913386412679516e-05,
+      "loss": 1.5199,
+      "step": 3610
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.913088771485974e-05,
+      "loss": 1.5251,
+      "step": 3620
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9127911302924324e-05,
+      "loss": 1.5158,
+      "step": 3630
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.912493489098892e-05,
+      "loss": 1.5103,
+      "step": 3640
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9121958479053504e-05,
+      "loss": 1.5286,
+      "step": 3650
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.911898206711809e-05,
+      "loss": 1.5275,
+      "step": 3660
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.911600565518268e-05,
+      "loss": 1.5083,
+      "step": 3670
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.911302924324727e-05,
+      "loss": 1.525,
+      "step": 3680
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.911005283131186e-05,
+      "loss": 1.5235,
+      "step": 3690
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9107076419376446e-05,
+      "loss": 1.5002,
+      "step": 3700
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.910410000744103e-05,
+      "loss": 1.5282,
+      "step": 3710
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.910112359550562e-05,
+      "loss": 1.5188,
+      "step": 3720
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.9098147183570206e-05,
+      "loss": 1.5043,
+      "step": 3730
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.909517077163479e-05,
+      "loss": 1.5069,
+      "step": 3740
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.909219435969939e-05,
+      "loss": 1.5304,
+      "step": 3750
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.9089217947763974e-05,
+      "loss": 1.5199,
+      "step": 3760
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.908624153582856e-05,
+      "loss": 1.5137,
+      "step": 3770
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.908326512389315e-05,
+      "loss": 1.5223,
+      "step": 3780
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.908028871195774e-05,
+      "loss": 1.517,
+      "step": 3790
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.907731230002233e-05,
+      "loss": 1.5034,
+      "step": 3800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.9074335888086915e-05,
+      "loss": 1.517,
+      "step": 3810
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.90713594761515e-05,
+      "loss": 1.5099,
+      "step": 3820
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.906838306421609e-05,
+      "loss": 1.498,
+      "step": 3830
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.9065406652280676e-05,
+      "loss": 1.5118,
+      "step": 3840
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.906243024034526e-05,
+      "loss": 1.4996,
+      "step": 3850
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.9059453828409856e-05,
+      "loss": 1.5113,
+      "step": 3860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.905647741647444e-05,
+      "loss": 1.494,
+      "step": 3870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.905350100453903e-05,
+      "loss": 1.5024,
+      "step": 3880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.905052459260362e-05,
+      "loss": 1.5118,
+      "step": 3890
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.904754818066821e-05,
+      "loss": 1.488,
+      "step": 3900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.90445717687328e-05,
+      "loss": 1.5065,
+      "step": 3910
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9041595356797385e-05,
+      "loss": 1.5061,
+      "step": 3920
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.903861894486197e-05,
+      "loss": 1.508,
+      "step": 3930
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.903564253292656e-05,
+      "loss": 1.4942,
+      "step": 3940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9032666120991145e-05,
+      "loss": 1.5033,
+      "step": 3950
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.902968970905574e-05,
+      "loss": 1.5048,
+      "step": 3960
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9026713297120326e-05,
+      "loss": 1.4964,
+      "step": 3970
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.902373688518491e-05,
+      "loss": 1.5118,
+      "step": 3980
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.90207604732495e-05,
+      "loss": 1.5213,
+      "step": 3990
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.901778406131409e-05,
+      "loss": 1.5256,
+      "step": 4000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.901480764937868e-05,
+      "loss": 1.4907,
+      "step": 4010
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.901183123744327e-05,
+      "loss": 1.5063,
+      "step": 4020
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9008854825507854e-05,
+      "loss": 1.5242,
+      "step": 4030
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.900587841357244e-05,
+      "loss": 1.4974,
+      "step": 4040
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.900290200163703e-05,
+      "loss": 1.5151,
+      "step": 4050
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.8999925589701615e-05,
+      "loss": 1.4985,
+      "step": 4060
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.899694917776621e-05,
+      "loss": 1.4966,
+      "step": 4070
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8993972765830795e-05,
+      "loss": 1.4988,
+      "step": 4080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.899099635389538e-05,
+      "loss": 1.4973,
+      "step": 4090
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.898801994195997e-05,
+      "loss": 1.498,
+      "step": 4100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.898504353002456e-05,
+      "loss": 1.4953,
+      "step": 4110
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.898206711808915e-05,
+      "loss": 1.5003,
+      "step": 4120
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.897909070615373e-05,
+      "loss": 1.5168,
+      "step": 4130
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.897611429421832e-05,
+      "loss": 1.5111,
+      "step": 4140
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.897313788228291e-05,
+      "loss": 1.498,
+      "step": 4150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.89701614703475e-05,
+      "loss": 1.4998,
+      "step": 4160
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.8967185058412084e-05,
+      "loss": 1.495,
+      "step": 4170
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.896420864647668e-05,
+      "loss": 1.4861,
+      "step": 4180
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.8961232234541265e-05,
+      "loss": 1.4928,
+      "step": 4190
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.895825582260585e-05,
+      "loss": 1.4995,
+      "step": 4200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.895527941067044e-05,
+      "loss": 1.4965,
+      "step": 4210
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.895230299873503e-05,
+      "loss": 1.4994,
+      "step": 4220
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.894932658679962e-05,
+      "loss": 1.4907,
+      "step": 4230
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8946350174864206e-05,
+      "loss": 1.5004,
+      "step": 4240
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.894337376292879e-05,
+      "loss": 1.4992,
+      "step": 4250
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.894039735099338e-05,
+      "loss": 1.503,
+      "step": 4260
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.893742093905797e-05,
+      "loss": 1.4895,
+      "step": 4270
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8934444527122554e-05,
+      "loss": 1.4848,
+      "step": 4280
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.893146811518715e-05,
+      "loss": 1.4815,
+      "step": 4290
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8928491703251734e-05,
+      "loss": 1.4897,
+      "step": 4300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.892551529131632e-05,
+      "loss": 1.4727,
+      "step": 4310
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.892253887938091e-05,
+      "loss": 1.4977,
+      "step": 4320
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.89195624674455e-05,
+      "loss": 1.5028,
+      "step": 4330
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.891658605551008e-05,
+      "loss": 1.4847,
+      "step": 4340
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8913609643574675e-05,
+      "loss": 1.4878,
+      "step": 4350
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.891063323163926e-05,
+      "loss": 1.478,
+      "step": 4360
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.890765681970385e-05,
+      "loss": 1.494,
+      "step": 4370
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8904680407768436e-05,
+      "loss": 1.4794,
+      "step": 4380
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.890170399583303e-05,
+      "loss": 1.4735,
+      "step": 4390
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.889872758389762e-05,
+      "loss": 1.4782,
+      "step": 4400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.8895751171962204e-05,
+      "loss": 1.4703,
+      "step": 4410
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.889277476002679e-05,
+      "loss": 1.5031,
+      "step": 4420
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.8889798348091384e-05,
+      "loss": 1.4813,
+      "step": 4430
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.8886821936155964e-05,
+      "loss": 1.4947,
+      "step": 4440
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.888384552422055e-05,
+      "loss": 1.4931,
+      "step": 4450
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.8880869112285145e-05,
+      "loss": 1.4838,
+      "step": 4460
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.887789270034973e-05,
+      "loss": 1.4984,
+      "step": 4470
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.887491628841432e-05,
+      "loss": 1.4859,
+      "step": 4480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8871939876478906e-05,
+      "loss": 1.4872,
+      "step": 4490
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.88689634645435e-05,
+      "loss": 1.4909,
+      "step": 4500
+    },
+    {
+      "epoch": 1.0,
+      "eval_cer": 4.69017094017094,
+      "eval_loss": 1.3148901462554932,
+      "eval_runtime": 5.6408,
+      "eval_samples_per_second": 1.773,
+      "eval_steps_per_second": 0.177,
+      "eval_wer": 1.1168831168831168,
+      "step": 4503
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8865987052608086e-05,
+      "loss": 1.4926,
+      "step": 4510
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.886301064067267e-05,
+      "loss": 1.4731,
+      "step": 4520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.886003422873726e-05,
+      "loss": 1.4858,
+      "step": 4530
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.885705781680185e-05,
+      "loss": 1.4486,
+      "step": 4540
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8854081404866434e-05,
+      "loss": 1.4782,
+      "step": 4550
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.885110499293102e-05,
+      "loss": 1.4846,
+      "step": 4560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8848128580995614e-05,
+      "loss": 1.47,
+      "step": 4570
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.88451521690602e-05,
+      "loss": 1.4743,
+      "step": 4580
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.884217575712479e-05,
+      "loss": 1.4752,
+      "step": 4590
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.8839199345189375e-05,
+      "loss": 1.475,
+      "step": 4600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.883622293325397e-05,
+      "loss": 1.4657,
+      "step": 4610
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.8833246521318556e-05,
+      "loss": 1.4593,
+      "step": 4620
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.883027010938314e-05,
+      "loss": 1.4578,
+      "step": 4630
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.882729369744773e-05,
+      "loss": 1.4776,
+      "step": 4640
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.8824317285512316e-05,
+      "loss": 1.4621,
+      "step": 4650
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.88213408735769e-05,
+      "loss": 1.4662,
+      "step": 4660
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.88183644616415e-05,
+      "loss": 1.4679,
+      "step": 4670
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.8815388049706084e-05,
+      "loss": 1.4585,
+      "step": 4680
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.881241163777067e-05,
+      "loss": 1.465,
+      "step": 4690
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.880943522583526e-05,
+      "loss": 1.4631,
+      "step": 4700
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.8806458813899844e-05,
+      "loss": 1.4629,
+      "step": 4710
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.880348240196444e-05,
+      "loss": 1.4545,
+      "step": 4720
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.8800505990029025e-05,
+      "loss": 1.459,
+      "step": 4730
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.879752957809361e-05,
+      "loss": 1.4657,
+      "step": 4740
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.87945531661582e-05,
+      "loss": 1.4664,
+      "step": 4750
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.8791576754222786e-05,
+      "loss": 1.444,
+      "step": 4760
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.878860034228737e-05,
+      "loss": 1.4447,
+      "step": 4770
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.8785623930351966e-05,
+      "loss": 1.4506,
+      "step": 4780
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.878264751841655e-05,
+      "loss": 1.4517,
+      "step": 4790
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.877967110648114e-05,
+      "loss": 1.4575,
+      "step": 4800
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.877669469454573e-05,
+      "loss": 1.469,
+      "step": 4810
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.877371828261032e-05,
+      "loss": 1.4533,
+      "step": 4820
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.877074187067491e-05,
+      "loss": 1.4464,
+      "step": 4830
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.8767765458739494e-05,
+      "loss": 1.4299,
+      "step": 4840
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.876478904680408e-05,
+      "loss": 1.4515,
+      "step": 4850
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.876181263486867e-05,
+      "loss": 1.4467,
+      "step": 4860
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.8758836222933255e-05,
+      "loss": 1.4511,
+      "step": 4870
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.875585981099784e-05,
+      "loss": 1.4352,
+      "step": 4880
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.8752883399062436e-05,
+      "loss": 1.4481,
+      "step": 4890
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.874990698712702e-05,
+      "loss": 1.4364,
+      "step": 4900
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.874693057519161e-05,
+      "loss": 1.4314,
+      "step": 4910
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.8743954163256196e-05,
+      "loss": 1.4184,
+      "step": 4920
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.874097775132079e-05,
+      "loss": 1.4295,
+      "step": 4930
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.873800133938538e-05,
+      "loss": 1.4238,
+      "step": 4940
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.873502492744996e-05,
+      "loss": 1.4238,
+      "step": 4950
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.873204851551455e-05,
+      "loss": 1.4335,
+      "step": 4960
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.872907210357914e-05,
+      "loss": 1.4314,
+      "step": 4970
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.8726095691643725e-05,
+      "loss": 1.419,
+      "step": 4980
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.872311927970831e-05,
+      "loss": 1.422,
+      "step": 4990
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.8720142867772905e-05,
+      "loss": 1.4268,
+      "step": 5000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.871716645583749e-05,
+      "loss": 1.4154,
+      "step": 5010
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.871419004390208e-05,
+      "loss": 1.4218,
+      "step": 5020
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.8711213631966666e-05,
+      "loss": 1.4336,
+      "step": 5030
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.870823722003126e-05,
+      "loss": 1.4207,
+      "step": 5040
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.8705260808095846e-05,
+      "loss": 1.4342,
+      "step": 5050
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.870228439616043e-05,
+      "loss": 1.4181,
+      "step": 5060
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.869930798422502e-05,
+      "loss": 1.4022,
+      "step": 5070
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.869633157228961e-05,
+      "loss": 1.4094,
+      "step": 5080
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.8693355160354194e-05,
+      "loss": 1.4184,
+      "step": 5090
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.869037874841879e-05,
+      "loss": 1.417,
+      "step": 5100
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.8687402336483375e-05,
+      "loss": 1.4027,
+      "step": 5110
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.868442592454796e-05,
+      "loss": 1.4239,
+      "step": 5120
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.868144951261255e-05,
+      "loss": 1.4125,
+      "step": 5130
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.8678473100677135e-05,
+      "loss": 1.4163,
+      "step": 5140
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.867549668874173e-05,
+      "loss": 1.4191,
+      "step": 5150
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.867252027680631e-05,
+      "loss": 1.4077,
+      "step": 5160
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.86695438648709e-05,
+      "loss": 1.4004,
+      "step": 5170
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.866656745293549e-05,
+      "loss": 1.4075,
+      "step": 5180
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.8663591041000077e-05,
+      "loss": 1.3929,
+      "step": 5190
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.8660614629064663e-05,
+      "loss": 1.4008,
+      "step": 5200
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.865763821712926e-05,
+      "loss": 1.3993,
+      "step": 5210
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.8654661805193844e-05,
+      "loss": 1.4076,
+      "step": 5220
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.865168539325843e-05,
+      "loss": 1.4128,
+      "step": 5230
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.864870898132302e-05,
+      "loss": 1.4082,
+      "step": 5240
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.864573256938761e-05,
+      "loss": 1.4015,
+      "step": 5250
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.864275615745219e-05,
+      "loss": 1.4157,
+      "step": 5260
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.863977974551678e-05,
+      "loss": 1.409,
+      "step": 5270
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.863680333358137e-05,
+      "loss": 1.4193,
+      "step": 5280
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.863382692164596e-05,
+      "loss": 1.4031,
+      "step": 5290
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.8630850509710546e-05,
+      "loss": 1.3987,
+      "step": 5300
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.862787409777513e-05,
+      "loss": 1.3973,
+      "step": 5310
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.8624897685839727e-05,
+      "loss": 1.3985,
+      "step": 5320
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.8621921273904313e-05,
+      "loss": 1.3926,
+      "step": 5330
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.86189448619689e-05,
+      "loss": 1.3898,
+      "step": 5340
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.861596845003349e-05,
+      "loss": 1.3939,
+      "step": 5350
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8612992038098074e-05,
+      "loss": 1.4008,
+      "step": 5360
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.861001562616266e-05,
+      "loss": 1.4226,
+      "step": 5370
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.860703921422725e-05,
+      "loss": 1.3995,
+      "step": 5380
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.860406280229184e-05,
+      "loss": 1.4043,
+      "step": 5390
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.860108639035643e-05,
+      "loss": 1.3828,
+      "step": 5400
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8598109978421015e-05,
+      "loss": 1.3749,
+      "step": 5410
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.85951335664856e-05,
+      "loss": 1.3977,
+      "step": 5420
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.8592157154550196e-05,
+      "loss": 1.4125,
+      "step": 5430
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.858918074261478e-05,
+      "loss": 1.4094,
+      "step": 5440
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.858620433067937e-05,
+      "loss": 1.3844,
+      "step": 5450
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.858322791874396e-05,
+      "loss": 1.3934,
+      "step": 5460
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.8580251506808544e-05,
+      "loss": 1.4037,
+      "step": 5470
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.857727509487313e-05,
+      "loss": 1.3893,
+      "step": 5480
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.8574298682937724e-05,
+      "loss": 1.389,
+      "step": 5490
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.857132227100231e-05,
+      "loss": 1.4043,
+      "step": 5500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.85683458590669e-05,
+      "loss": 1.3974,
+      "step": 5510
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.8565369447131485e-05,
+      "loss": 1.3681,
+      "step": 5520
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.856239303519608e-05,
+      "loss": 1.3802,
+      "step": 5530
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.8559416623260665e-05,
+      "loss": 1.3917,
+      "step": 5540
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.855644021132525e-05,
+      "loss": 1.3793,
+      "step": 5550
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.855346379938984e-05,
+      "loss": 1.3902,
+      "step": 5560
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.8550487387454426e-05,
+      "loss": 1.3861,
+      "step": 5570
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.854751097551901e-05,
+      "loss": 1.3954,
+      "step": 5580
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.85445345635836e-05,
+      "loss": 1.3891,
+      "step": 5590
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.8541558151648194e-05,
+      "loss": 1.3867,
+      "step": 5600
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.853858173971278e-05,
+      "loss": 1.3819,
+      "step": 5610
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.853560532777737e-05,
+      "loss": 1.3841,
+      "step": 5620
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.8532628915841954e-05,
+      "loss": 1.369,
+      "step": 5630
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.852965250390655e-05,
+      "loss": 1.3736,
+      "step": 5640
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.8526676091971135e-05,
+      "loss": 1.3788,
+      "step": 5650
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.852369968003572e-05,
+      "loss": 1.3843,
+      "step": 5660
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.852072326810031e-05,
+      "loss": 1.3877,
+      "step": 5670
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.8517746856164896e-05,
+      "loss": 1.382,
+      "step": 5680
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.851477044422948e-05,
+      "loss": 1.3705,
+      "step": 5690
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.851179403229407e-05,
+      "loss": 1.3871,
+      "step": 5700
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.850881762035866e-05,
+      "loss": 1.3757,
+      "step": 5710
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.850584120842325e-05,
+      "loss": 1.3762,
+      "step": 5720
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.850286479648784e-05,
+      "loss": 1.3688,
+      "step": 5730
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.8499888384552424e-05,
+      "loss": 1.3845,
+      "step": 5740
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.849691197261702e-05,
+      "loss": 1.3674,
+      "step": 5750
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.8493935560681604e-05,
+      "loss": 1.3719,
+      "step": 5760
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.8490959148746184e-05,
+      "loss": 1.3572,
+      "step": 5770
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.848798273681078e-05,
+      "loss": 1.3708,
+      "step": 5780
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.8485006324875365e-05,
+      "loss": 1.3611,
+      "step": 5790
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.848202991293995e-05,
+      "loss": 1.3665,
+      "step": 5800
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.847905350100454e-05,
+      "loss": 1.3625,
+      "step": 5810
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.847607708906913e-05,
+      "loss": 1.3655,
+      "step": 5820
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.847310067713372e-05,
+      "loss": 1.3733,
+      "step": 5830
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.8470124265198306e-05,
+      "loss": 1.364,
+      "step": 5840
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.846714785326289e-05,
+      "loss": 1.3572,
+      "step": 5850
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.846417144132749e-05,
+      "loss": 1.3562,
+      "step": 5860
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.846119502939207e-05,
+      "loss": 1.3629,
+      "step": 5870
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.845821861745666e-05,
+      "loss": 1.3697,
+      "step": 5880
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.845524220552125e-05,
+      "loss": 1.3615,
+      "step": 5890
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.8452265793585834e-05,
+      "loss": 1.3758,
+      "step": 5900
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.844928938165042e-05,
+      "loss": 1.3569,
+      "step": 5910
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.8446312969715015e-05,
+      "loss": 1.3677,
+      "step": 5920
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.84433365577796e-05,
+      "loss": 1.376,
+      "step": 5930
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.844036014584419e-05,
+      "loss": 1.3629,
+      "step": 5940
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.8437383733908776e-05,
+      "loss": 1.3652,
+      "step": 5950
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.843440732197337e-05,
+      "loss": 1.3707,
+      "step": 5960
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.8431430910037956e-05,
+      "loss": 1.3635,
+      "step": 5970
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.8428454498102536e-05,
+      "loss": 1.3577,
+      "step": 5980
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.842547808616713e-05,
+      "loss": 1.3464,
+      "step": 5990
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.842250167423172e-05,
+      "loss": 1.3467,
+      "step": 6000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.8419525262296304e-05,
+      "loss": 1.3635,
+      "step": 6010
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.841654885036089e-05,
+      "loss": 1.3531,
+      "step": 6020
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.8413572438425484e-05,
+      "loss": 1.3516,
+      "step": 6030
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.841059602649007e-05,
+      "loss": 1.3668,
+      "step": 6040
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.840761961455466e-05,
+      "loss": 1.3615,
+      "step": 6050
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.8404643202619245e-05,
+      "loss": 1.3707,
+      "step": 6060
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.840166679068384e-05,
+      "loss": 1.3546,
+      "step": 6070
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.839869037874842e-05,
+      "loss": 1.355,
+      "step": 6080
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.8395713966813006e-05,
+      "loss": 1.3625,
+      "step": 6090
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.83927375548776e-05,
+      "loss": 1.3644,
+      "step": 6100
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.8389761142942186e-05,
+      "loss": 1.3525,
+      "step": 6110
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.838678473100677e-05,
+      "loss": 1.3383,
+      "step": 6120
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.838380831907136e-05,
+      "loss": 1.3394,
+      "step": 6130
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.8380831907135954e-05,
+      "loss": 1.3548,
+      "step": 6140
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.837785549520054e-05,
+      "loss": 1.3565,
+      "step": 6150
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.837487908326513e-05,
+      "loss": 1.3586,
+      "step": 6160
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.8371902671329715e-05,
+      "loss": 1.3674,
+      "step": 6170
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.83689262593943e-05,
+      "loss": 1.3495,
+      "step": 6180
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.836594984745889e-05,
+      "loss": 1.3432,
+      "step": 6190
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.8362973435523475e-05,
+      "loss": 1.349,
+      "step": 6200
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.835999702358807e-05,
+      "loss": 1.354,
+      "step": 6210
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.8357020611652656e-05,
+      "loss": 1.3522,
+      "step": 6220
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.835404419971724e-05,
+      "loss": 1.3609,
+      "step": 6230
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.835106778778183e-05,
+      "loss": 1.3296,
+      "step": 6240
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.834809137584642e-05,
+      "loss": 1.3521,
+      "step": 6250
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.834511496391101e-05,
+      "loss": 1.3495,
+      "step": 6260
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.83421385519756e-05,
+      "loss": 1.3502,
+      "step": 6270
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.8339162140040184e-05,
+      "loss": 1.3451,
+      "step": 6280
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.833618572810477e-05,
+      "loss": 1.3381,
+      "step": 6290
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.833320931616936e-05,
+      "loss": 1.335,
+      "step": 6300
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.833023290423395e-05,
+      "loss": 1.3475,
+      "step": 6310
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.832725649229854e-05,
+      "loss": 1.3534,
+      "step": 6320
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.8324280080363125e-05,
+      "loss": 1.349,
+      "step": 6330
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.832130366842771e-05,
+      "loss": 1.3437,
+      "step": 6340
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.8318327256492306e-05,
+      "loss": 1.3464,
+      "step": 6350
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.831535084455689e-05,
+      "loss": 1.3517,
+      "step": 6360
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.831237443262148e-05,
+      "loss": 1.348,
+      "step": 6370
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.8309398020686067e-05,
+      "loss": 1.3504,
+      "step": 6380
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.8306421608750653e-05,
+      "loss": 1.3401,
+      "step": 6390
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.830344519681524e-05,
+      "loss": 1.3453,
+      "step": 6400
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.830046878487983e-05,
+      "loss": 1.34,
+      "step": 6410
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.829749237294442e-05,
+      "loss": 1.3435,
+      "step": 6420
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.829451596100901e-05,
+      "loss": 1.3427,
+      "step": 6430
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.8291539549073595e-05,
+      "loss": 1.3379,
+      "step": 6440
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.828856313713818e-05,
+      "loss": 1.3315,
+      "step": 6450
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.8285586725202775e-05,
+      "loss": 1.3286,
+      "step": 6460
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.828261031326736e-05,
+      "loss": 1.3378,
+      "step": 6470
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.827963390133195e-05,
+      "loss": 1.3354,
+      "step": 6480
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8276657489396536e-05,
+      "loss": 1.3238,
+      "step": 6490
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.827368107746112e-05,
+      "loss": 1.334,
+      "step": 6500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.827070466552571e-05,
+      "loss": 1.3356,
+      "step": 6510
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.82677282535903e-05,
+      "loss": 1.3249,
+      "step": 6520
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.826475184165489e-05,
+      "loss": 1.3282,
+      "step": 6530
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.826177542971948e-05,
+      "loss": 1.3448,
+      "step": 6540
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.8258799017784064e-05,
+      "loss": 1.3322,
+      "step": 6550
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.825582260584865e-05,
+      "loss": 1.3375,
+      "step": 6560
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.8252846193913245e-05,
+      "loss": 1.3331,
+      "step": 6570
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.824986978197783e-05,
+      "loss": 1.3254,
+      "step": 6580
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.824689337004241e-05,
+      "loss": 1.3414,
+      "step": 6590
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.8243916958107005e-05,
+      "loss": 1.3346,
+      "step": 6600
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.824094054617159e-05,
+      "loss": 1.3497,
+      "step": 6610
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.823796413423618e-05,
+      "loss": 1.3214,
+      "step": 6620
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.8234987722300766e-05,
+      "loss": 1.3413,
+      "step": 6630
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.823201131036536e-05,
+      "loss": 1.3294,
+      "step": 6640
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.822903489842995e-05,
+      "loss": 1.3275,
+      "step": 6650
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.8226058486494534e-05,
+      "loss": 1.3325,
+      "step": 6660
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.822308207455912e-05,
+      "loss": 1.3388,
+      "step": 6670
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.8220105662623714e-05,
+      "loss": 1.3361,
+      "step": 6680
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.8217129250688294e-05,
+      "loss": 1.3263,
+      "step": 6690
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.821415283875289e-05,
+      "loss": 1.333,
+      "step": 6700
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.8211176426817475e-05,
+      "loss": 1.3163,
+      "step": 6710
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.820820001488206e-05,
+      "loss": 1.3371,
+      "step": 6720
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.820522360294665e-05,
+      "loss": 1.3405,
+      "step": 6730
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.820224719101124e-05,
+      "loss": 1.3207,
+      "step": 6740
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.819927077907583e-05,
+      "loss": 1.3358,
+      "step": 6750
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.8196294367140416e-05,
+      "loss": 1.3252,
+      "step": 6760
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.8193317955205e-05,
+      "loss": 1.3308,
+      "step": 6770
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.81903415432696e-05,
+      "loss": 1.3392,
+      "step": 6780
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.8187365131334184e-05,
+      "loss": 1.3205,
+      "step": 6790
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.8184388719398764e-05,
+      "loss": 1.3354,
+      "step": 6800
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.818141230746336e-05,
+      "loss": 1.3209,
+      "step": 6810
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.8178435895527944e-05,
+      "loss": 1.3408,
+      "step": 6820
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.817545948359253e-05,
+      "loss": 1.3272,
+      "step": 6830
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.817248307165712e-05,
+      "loss": 1.3233,
+      "step": 6840
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.816950665972171e-05,
+      "loss": 1.3371,
+      "step": 6850
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.81665302477863e-05,
+      "loss": 1.339,
+      "step": 6860
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.8163553835850886e-05,
+      "loss": 1.3347,
+      "step": 6870
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.816057742391547e-05,
+      "loss": 1.3167,
+      "step": 6880
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.8157601011980066e-05,
+      "loss": 1.3208,
+      "step": 6890
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.8154624600044646e-05,
+      "loss": 1.3216,
+      "step": 6900
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.815164818810923e-05,
+      "loss": 1.323,
+      "step": 6910
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.814867177617383e-05,
+      "loss": 1.3258,
+      "step": 6920
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.8145695364238414e-05,
+      "loss": 1.3157,
+      "step": 6930
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.8142718952303e-05,
+      "loss": 1.3214,
+      "step": 6940
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.813974254036759e-05,
+      "loss": 1.3354,
+      "step": 6950
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.813676612843218e-05,
+      "loss": 1.3185,
+      "step": 6960
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.813378971649677e-05,
+      "loss": 1.3367,
+      "step": 6970
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.8130813304561355e-05,
+      "loss": 1.3354,
+      "step": 6980
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.812783689262594e-05,
+      "loss": 1.3301,
+      "step": 6990
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.812486048069053e-05,
+      "loss": 1.3191,
+      "step": 7000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.8121884068755116e-05,
+      "loss": 1.3136,
+      "step": 7010
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.81189076568197e-05,
+      "loss": 1.3258,
+      "step": 7020
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.8115931244884296e-05,
+      "loss": 1.3247,
+      "step": 7030
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.811295483294888e-05,
+      "loss": 1.3105,
+      "step": 7040
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.810997842101347e-05,
+      "loss": 1.3231,
+      "step": 7050
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.810700200907806e-05,
+      "loss": 1.3237,
+      "step": 7060
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.810402559714265e-05,
+      "loss": 1.3178,
+      "step": 7070
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.810104918520724e-05,
+      "loss": 1.317,
+      "step": 7080
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.8098072773271824e-05,
+      "loss": 1.3293,
+      "step": 7090
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.809509636133641e-05,
+      "loss": 1.3291,
+      "step": 7100
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.8092119949401e-05,
+      "loss": 1.3165,
+      "step": 7110
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.8089143537465585e-05,
+      "loss": 1.3253,
+      "step": 7120
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.808616712553018e-05,
+      "loss": 1.3076,
+      "step": 7130
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.8083190713594766e-05,
+      "loss": 1.3237,
+      "step": 7140
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.808021430165935e-05,
+      "loss": 1.3155,
+      "step": 7150
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.807723788972394e-05,
+      "loss": 1.3141,
+      "step": 7160
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.807426147778853e-05,
+      "loss": 1.3134,
+      "step": 7170
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.807128506585312e-05,
+      "loss": 1.3274,
+      "step": 7180
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.806830865391771e-05,
+      "loss": 1.309,
+      "step": 7190
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.8065332241982294e-05,
+      "loss": 1.3044,
+      "step": 7200
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.806235583004688e-05,
+      "loss": 1.3058,
+      "step": 7210
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.805937941811147e-05,
+      "loss": 1.3185,
+      "step": 7220
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.8056403006176055e-05,
+      "loss": 1.3152,
+      "step": 7230
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.805342659424065e-05,
+      "loss": 1.3186,
+      "step": 7240
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.8050450182305235e-05,
+      "loss": 1.3153,
+      "step": 7250
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.804747377036982e-05,
+      "loss": 1.3061,
+      "step": 7260
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.804449735843441e-05,
+      "loss": 1.3144,
+      "step": 7270
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.8041520946499e-05,
+      "loss": 1.3165,
+      "step": 7280
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.803854453456359e-05,
+      "loss": 1.3091,
+      "step": 7290
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.8035568122628176e-05,
+      "loss": 1.3199,
+      "step": 7300
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.803259171069276e-05,
+      "loss": 1.3201,
+      "step": 7310
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.802961529875735e-05,
+      "loss": 1.3095,
+      "step": 7320
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.802663888682194e-05,
+      "loss": 1.3148,
+      "step": 7330
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.8023662474886524e-05,
+      "loss": 1.3008,
+      "step": 7340
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.802068606295112e-05,
+      "loss": 1.3027,
+      "step": 7350
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.8017709651015705e-05,
+      "loss": 1.3095,
+      "step": 7360
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.801473323908029e-05,
+      "loss": 1.3031,
+      "step": 7370
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.801175682714488e-05,
+      "loss": 1.3165,
+      "step": 7380
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.800878041520947e-05,
+      "loss": 1.3206,
+      "step": 7390
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.800580400327406e-05,
+      "loss": 1.3076,
+      "step": 7400
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.800282759133864e-05,
+      "loss": 1.3143,
+      "step": 7410
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.799985117940323e-05,
+      "loss": 1.3227,
+      "step": 7420
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.799687476746782e-05,
+      "loss": 1.3159,
+      "step": 7430
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.7993898355532407e-05,
+      "loss": 1.2961,
+      "step": 7440
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.7990921943596993e-05,
+      "loss": 1.3064,
+      "step": 7450
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.798794553166159e-05,
+      "loss": 1.309,
+      "step": 7460
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.7984969119726174e-05,
+      "loss": 1.3039,
+      "step": 7470
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.798199270779076e-05,
+      "loss": 1.3091,
+      "step": 7480
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.797901629585535e-05,
+      "loss": 1.299,
+      "step": 7490
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.797603988391994e-05,
+      "loss": 1.302,
+      "step": 7500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.797306347198452e-05,
+      "loss": 1.3035,
+      "step": 7510
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.7970087060049115e-05,
+      "loss": 1.3283,
+      "step": 7520
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.79671106481137e-05,
+      "loss": 1.3098,
+      "step": 7530
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.796413423617829e-05,
+      "loss": 1.3179,
+      "step": 7540
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.7961157824242876e-05,
+      "loss": 1.3247,
+      "step": 7550
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.795818141230747e-05,
+      "loss": 1.2883,
+      "step": 7560
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.7955205000372057e-05,
+      "loss": 1.296,
+      "step": 7570
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.7952228588436644e-05,
+      "loss": 1.3131,
+      "step": 7580
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.794925217650123e-05,
+      "loss": 1.3031,
+      "step": 7590
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.7946275764565824e-05,
+      "loss": 1.3067,
+      "step": 7600
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.7943299352630404e-05,
+      "loss": 1.2966,
+      "step": 7610
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.794032294069499e-05,
+      "loss": 1.3065,
+      "step": 7620
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.7937346528759585e-05,
+      "loss": 1.3114,
+      "step": 7630
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.793437011682417e-05,
+      "loss": 1.2946,
+      "step": 7640
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.793139370488876e-05,
+      "loss": 1.3103,
+      "step": 7650
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.7928417292953345e-05,
+      "loss": 1.2927,
+      "step": 7660
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.792544088101794e-05,
+      "loss": 1.3005,
+      "step": 7670
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.7922464469082526e-05,
+      "loss": 1.2924,
+      "step": 7680
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.791948805714711e-05,
+      "loss": 1.3107,
+      "step": 7690
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.79165116452117e-05,
+      "loss": 1.3135,
+      "step": 7700
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.7913535233276294e-05,
+      "loss": 1.3041,
+      "step": 7710
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.7910558821340874e-05,
+      "loss": 1.3062,
+      "step": 7720
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.790758240940546e-05,
+      "loss": 1.3002,
+      "step": 7730
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.7904605997470054e-05,
+      "loss": 1.2952,
+      "step": 7740
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.790162958553464e-05,
+      "loss": 1.2926,
+      "step": 7750
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.789865317359923e-05,
+      "loss": 1.3107,
+      "step": 7760
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.7895676761663815e-05,
+      "loss": 1.3047,
+      "step": 7770
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.789270034972841e-05,
+      "loss": 1.3062,
+      "step": 7780
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.7889723937792995e-05,
+      "loss": 1.2938,
+      "step": 7790
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.788674752585758e-05,
+      "loss": 1.3065,
+      "step": 7800
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.788377111392217e-05,
+      "loss": 1.2996,
+      "step": 7810
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7880794701986756e-05,
+      "loss": 1.2945,
+      "step": 7820
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.787781829005134e-05,
+      "loss": 1.3094,
+      "step": 7830
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.787484187811593e-05,
+      "loss": 1.2981,
+      "step": 7840
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7871865466180524e-05,
+      "loss": 1.3001,
+      "step": 7850
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.786888905424511e-05,
+      "loss": 1.3054,
+      "step": 7860
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.78659126423097e-05,
+      "loss": 1.2942,
+      "step": 7870
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7862936230374284e-05,
+      "loss": 1.2877,
+      "step": 7880
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.785995981843888e-05,
+      "loss": 1.3055,
+      "step": 7890
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7856983406503465e-05,
+      "loss": 1.2837,
+      "step": 7900
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.785400699456805e-05,
+      "loss": 1.2919,
+      "step": 7910
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.785103058263264e-05,
+      "loss": 1.2827,
+      "step": 7920
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7848054170697226e-05,
+      "loss": 1.2806,
+      "step": 7930
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.784507775876181e-05,
+      "loss": 1.2827,
+      "step": 7940
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7842101346826406e-05,
+      "loss": 1.2921,
+      "step": 7950
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.783912493489099e-05,
+      "loss": 1.2928,
+      "step": 7960
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.783614852295558e-05,
+      "loss": 1.3016,
+      "step": 7970
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.783317211102017e-05,
+      "loss": 1.2841,
+      "step": 7980
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.783019569908476e-05,
+      "loss": 1.3047,
+      "step": 7990
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.782721928714935e-05,
+      "loss": 1.2971,
+      "step": 8000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7824242875213934e-05,
+      "loss": 1.2813,
+      "step": 8010
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.782126646327852e-05,
+      "loss": 1.2758,
+      "step": 8020
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.781829005134311e-05,
+      "loss": 1.2962,
+      "step": 8030
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7815313639407695e-05,
+      "loss": 1.287,
+      "step": 8040
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.781233722747228e-05,
+      "loss": 1.3027,
+      "step": 8050
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7809360815536876e-05,
+      "loss": 1.2932,
+      "step": 8060
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.780638440360146e-05,
+      "loss": 1.2765,
+      "step": 8070
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.780340799166605e-05,
+      "loss": 1.2905,
+      "step": 8080
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7800431579730636e-05,
+      "loss": 1.2983,
+      "step": 8090
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.779745516779523e-05,
+      "loss": 1.2811,
+      "step": 8100
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.779447875585982e-05,
+      "loss": 1.2961,
+      "step": 8110
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7791502343924404e-05,
+      "loss": 1.2794,
+      "step": 8120
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.778852593198899e-05,
+      "loss": 1.287,
+      "step": 8130
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.778554952005358e-05,
+      "loss": 1.3107,
+      "step": 8140
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.7782573108118165e-05,
+      "loss": 1.2854,
+      "step": 8150
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.777959669618275e-05,
+      "loss": 1.2832,
+      "step": 8160
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.7776620284247345e-05,
+      "loss": 1.2763,
+      "step": 8170
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.777364387231193e-05,
+      "loss": 1.2793,
+      "step": 8180
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.777066746037652e-05,
+      "loss": 1.2963,
+      "step": 8190
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.7767691048441106e-05,
+      "loss": 1.2826,
+      "step": 8200
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.77647146365057e-05,
+      "loss": 1.2796,
+      "step": 8210
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7761738224570286e-05,
+      "loss": 1.2846,
+      "step": 8220
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7758761812634866e-05,
+      "loss": 1.2826,
+      "step": 8230
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.775578540069946e-05,
+      "loss": 1.2809,
+      "step": 8240
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.775280898876405e-05,
+      "loss": 1.2859,
+      "step": 8250
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7749832576828634e-05,
+      "loss": 1.2897,
+      "step": 8260
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.774685616489322e-05,
+      "loss": 1.2835,
+      "step": 8270
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.7743879752957815e-05,
+      "loss": 1.2798,
+      "step": 8280
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.77409033410224e-05,
+      "loss": 1.2706,
+      "step": 8290
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.773792692908699e-05,
+      "loss": 1.2853,
+      "step": 8300
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.7734950517151575e-05,
+      "loss": 1.2937,
+      "step": 8310
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.773197410521617e-05,
+      "loss": 1.2922,
+      "step": 8320
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.772899769328075e-05,
+      "loss": 1.2876,
+      "step": 8330
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.772602128134534e-05,
+      "loss": 1.2746,
+      "step": 8340
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.772304486940993e-05,
+      "loss": 1.2877,
+      "step": 8350
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7720068457474516e-05,
+      "loss": 1.2732,
+      "step": 8360
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.77170920455391e-05,
+      "loss": 1.2755,
+      "step": 8370
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.77141156336037e-05,
+      "loss": 1.2722,
+      "step": 8380
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7711139221668284e-05,
+      "loss": 1.2914,
+      "step": 8390
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.770816280973287e-05,
+      "loss": 1.2845,
+      "step": 8400
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.770518639779746e-05,
+      "loss": 1.2751,
+      "step": 8410
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.770220998586205e-05,
+      "loss": 1.2764,
+      "step": 8420
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.769923357392663e-05,
+      "loss": 1.2789,
+      "step": 8430
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.769625716199122e-05,
+      "loss": 1.2916,
+      "step": 8440
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.769328075005581e-05,
+      "loss": 1.2739,
+      "step": 8450
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.76903043381204e-05,
+      "loss": 1.2705,
+      "step": 8460
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.7687327926184986e-05,
+      "loss": 1.2793,
+      "step": 8470
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.768435151424957e-05,
+      "loss": 1.2734,
+      "step": 8480
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.7681375102314166e-05,
+      "loss": 1.2813,
+      "step": 8490
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.7678398690378753e-05,
+      "loss": 1.2794,
+      "step": 8500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.767542227844334e-05,
+      "loss": 1.278,
+      "step": 8510
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.767244586650793e-05,
+      "loss": 1.2579,
+      "step": 8520
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.766946945457252e-05,
+      "loss": 1.2898,
+      "step": 8530
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.76664930426371e-05,
+      "loss": 1.2864,
+      "step": 8540
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.766351663070169e-05,
+      "loss": 1.2807,
+      "step": 8550
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.766054021876628e-05,
+      "loss": 1.2759,
+      "step": 8560
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.765756380683087e-05,
+      "loss": 1.2835,
+      "step": 8570
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.7654587394895455e-05,
+      "loss": 1.2992,
+      "step": 8580
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.765161098296004e-05,
+      "loss": 1.2739,
+      "step": 8590
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.7648634571024636e-05,
+      "loss": 1.2866,
+      "step": 8600
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.764565815908922e-05,
+      "loss": 1.2804,
+      "step": 8610
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.764268174715381e-05,
+      "loss": 1.2865,
+      "step": 8620
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.76397053352184e-05,
+      "loss": 1.2904,
+      "step": 8630
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.7636728923282984e-05,
+      "loss": 1.2872,
+      "step": 8640
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.763375251134757e-05,
+      "loss": 1.2725,
+      "step": 8650
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.763077609941216e-05,
+      "loss": 1.2874,
+      "step": 8660
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.762779968747675e-05,
+      "loss": 1.2714,
+      "step": 8670
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.762482327554134e-05,
+      "loss": 1.2699,
+      "step": 8680
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.7621846863605925e-05,
+      "loss": 1.2727,
+      "step": 8690
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.761887045167051e-05,
+      "loss": 1.2732,
+      "step": 8700
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.7615894039735105e-05,
+      "loss": 1.2711,
+      "step": 8710
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.761291762779969e-05,
+      "loss": 1.267,
+      "step": 8720
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.760994121586428e-05,
+      "loss": 1.2725,
+      "step": 8730
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.7606964803928866e-05,
+      "loss": 1.273,
+      "step": 8740
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.760398839199345e-05,
+      "loss": 1.2794,
+      "step": 8750
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.760101198005804e-05,
+      "loss": 1.279,
+      "step": 8760
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7598035568122634e-05,
+      "loss": 1.2688,
+      "step": 8770
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.759505915618722e-05,
+      "loss": 1.2731,
+      "step": 8780
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.759208274425181e-05,
+      "loss": 1.2664,
+      "step": 8790
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7589106332316394e-05,
+      "loss": 1.2708,
+      "step": 8800
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.758612992038099e-05,
+      "loss": 1.2672,
+      "step": 8810
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.7583153508445575e-05,
+      "loss": 1.2617,
+      "step": 8820
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.758017709651016e-05,
+      "loss": 1.2721,
+      "step": 8830
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.757720068457475e-05,
+      "loss": 1.2667,
+      "step": 8840
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.7574224272639336e-05,
+      "loss": 1.2706,
+      "step": 8850
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.757124786070392e-05,
+      "loss": 1.2882,
+      "step": 8860
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.756827144876851e-05,
+      "loss": 1.2705,
+      "step": 8870
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.75652950368331e-05,
+      "loss": 1.2737,
+      "step": 8880
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.756231862489769e-05,
+      "loss": 1.2712,
+      "step": 8890
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.755934221296228e-05,
+      "loss": 1.2758,
+      "step": 8900
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.7556365801026864e-05,
+      "loss": 1.2616,
+      "step": 8910
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.755338938909146e-05,
+      "loss": 1.2658,
+      "step": 8920
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.7550412977156044e-05,
+      "loss": 1.2787,
+      "step": 8930
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7547436565220624e-05,
+      "loss": 1.2606,
+      "step": 8940
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.754446015328522e-05,
+      "loss": 1.2642,
+      "step": 8950
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7541483741349805e-05,
+      "loss": 1.2595,
+      "step": 8960
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.753850732941439e-05,
+      "loss": 1.2627,
+      "step": 8970
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.753553091747898e-05,
+      "loss": 1.2674,
+      "step": 8980
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.753255450554357e-05,
+      "loss": 1.2737,
+      "step": 8990
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.752957809360816e-05,
+      "loss": 1.2818,
+      "step": 9000
+    },
+    {
+      "epoch": 2.0,
+      "eval_cer": 4.782051282051282,
+      "eval_loss": 1.1977766752243042,
+      "eval_runtime": 5.6706,
+      "eval_samples_per_second": 1.763,
+      "eval_steps_per_second": 0.176,
+      "eval_wer": 1.0389610389610389,
+      "step": 9006
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.7526601681672746e-05,
+      "loss": 1.2688,
+      "step": 9010
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.752362526973733e-05,
+      "loss": 1.2758,
+      "step": 9020
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.752064885780193e-05,
+      "loss": 1.2735,
+      "step": 9030
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.7517672445866514e-05,
+      "loss": 1.2591,
+      "step": 9040
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.75146960339311e-05,
+      "loss": 1.2773,
+      "step": 9050
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.751171962199569e-05,
+      "loss": 1.263,
+      "step": 9060
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.7508743210060274e-05,
+      "loss": 1.2699,
+      "step": 9070
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.750576679812486e-05,
+      "loss": 1.2631,
+      "step": 9080
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.750279038618945e-05,
+      "loss": 1.2583,
+      "step": 9090
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.749981397425404e-05,
+      "loss": 1.2732,
+      "step": 9100
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.749683756231863e-05,
+      "loss": 1.2596,
+      "step": 9110
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.7493861150383216e-05,
+      "loss": 1.2627,
+      "step": 9120
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.74908847384478e-05,
+      "loss": 1.273,
+      "step": 9130
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.7487908326512396e-05,
+      "loss": 1.2677,
+      "step": 9140
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.7484931914576976e-05,
+      "loss": 1.2619,
+      "step": 9150
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.748195550264157e-05,
+      "loss": 1.2703,
+      "step": 9160
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.747897909070616e-05,
+      "loss": 1.2778,
+      "step": 9170
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.7476002678770744e-05,
+      "loss": 1.2628,
+      "step": 9180
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.747302626683533e-05,
+      "loss": 1.2509,
+      "step": 9190
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.7470049854899924e-05,
+      "loss": 1.2606,
+      "step": 9200
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.746707344296451e-05,
+      "loss": 1.266,
+      "step": 9210
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.74640970310291e-05,
+      "loss": 1.2697,
+      "step": 9220
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.7461120619093685e-05,
+      "loss": 1.2683,
+      "step": 9230
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.745814420715828e-05,
+      "loss": 1.2571,
+      "step": 9240
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.745516779522286e-05,
+      "loss": 1.2676,
+      "step": 9250
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.7452191383287446e-05,
+      "loss": 1.2674,
+      "step": 9260
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.744921497135204e-05,
+      "loss": 1.2688,
+      "step": 9270
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.7446238559416626e-05,
+      "loss": 1.2669,
+      "step": 9280
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.744326214748121e-05,
+      "loss": 1.2627,
+      "step": 9290
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.74402857355458e-05,
+      "loss": 1.2678,
+      "step": 9300
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.7437309323610394e-05,
+      "loss": 1.2664,
+      "step": 9310
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.743433291167498e-05,
+      "loss": 1.2744,
+      "step": 9320
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.743135649973957e-05,
+      "loss": 1.2556,
+      "step": 9330
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.7428380087804155e-05,
+      "loss": 1.2537,
+      "step": 9340
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.742540367586874e-05,
+      "loss": 1.2616,
+      "step": 9350
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.742242726393333e-05,
+      "loss": 1.2558,
+      "step": 9360
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.7419450851997915e-05,
+      "loss": 1.2721,
+      "step": 9370
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.741647444006251e-05,
+      "loss": 1.2562,
+      "step": 9380
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.7413498028127096e-05,
+      "loss": 1.2503,
+      "step": 9390
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.741052161619168e-05,
+      "loss": 1.2612,
+      "step": 9400
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.740754520425627e-05,
+      "loss": 1.2612,
+      "step": 9410
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.740456879232086e-05,
+      "loss": 1.2567,
+      "step": 9420
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.740159238038545e-05,
+      "loss": 1.2588,
+      "step": 9430
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.739861596845004e-05,
+      "loss": 1.2572,
+      "step": 9440
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.7395639556514624e-05,
+      "loss": 1.2568,
+      "step": 9450
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.739266314457921e-05,
+      "loss": 1.2582,
+      "step": 9460
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.73896867326438e-05,
+      "loss": 1.2638,
+      "step": 9470
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.738671032070839e-05,
+      "loss": 1.2651,
+      "step": 9480
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.738373390877298e-05,
+      "loss": 1.2555,
+      "step": 9490
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.7380757496837565e-05,
+      "loss": 1.2537,
+      "step": 9500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.737778108490215e-05,
+      "loss": 1.2474,
+      "step": 9510
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.737480467296674e-05,
+      "loss": 1.2725,
+      "step": 9520
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.737182826103133e-05,
+      "loss": 1.2583,
+      "step": 9530
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.736885184909592e-05,
+      "loss": 1.2605,
+      "step": 9540
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.7365875437160507e-05,
+      "loss": 1.2681,
+      "step": 9550
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.7362899025225093e-05,
+      "loss": 1.2609,
+      "step": 9560
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.735992261328968e-05,
+      "loss": 1.2714,
+      "step": 9570
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.735694620135427e-05,
+      "loss": 1.261,
+      "step": 9580
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.735396978941886e-05,
+      "loss": 1.2623,
+      "step": 9590
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.735099337748345e-05,
+      "loss": 1.2561,
+      "step": 9600
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.7348016965548035e-05,
+      "loss": 1.2412,
+      "step": 9610
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.734504055361262e-05,
+      "loss": 1.2623,
+      "step": 9620
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.7342064141677215e-05,
+      "loss": 1.2684,
+      "step": 9630
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.73390877297418e-05,
+      "loss": 1.2627,
+      "step": 9640
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.733611131780639e-05,
+      "loss": 1.2657,
+      "step": 9650
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.7333134905870976e-05,
+      "loss": 1.2496,
+      "step": 9660
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.733015849393556e-05,
+      "loss": 1.2617,
+      "step": 9670
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.732718208200015e-05,
+      "loss": 1.2403,
+      "step": 9680
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.732420567006474e-05,
+      "loss": 1.2727,
+      "step": 9690
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.732122925812933e-05,
+      "loss": 1.2525,
+      "step": 9700
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.731825284619392e-05,
+      "loss": 1.2576,
+      "step": 9710
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7315276434258504e-05,
+      "loss": 1.2627,
+      "step": 9720
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.731230002232309e-05,
+      "loss": 1.2637,
+      "step": 9730
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7309323610387685e-05,
+      "loss": 1.2594,
+      "step": 9740
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.730634719845227e-05,
+      "loss": 1.2494,
+      "step": 9750
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.730337078651685e-05,
+      "loss": 1.2612,
+      "step": 9760
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7300394374581445e-05,
+      "loss": 1.2578,
+      "step": 9770
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.729741796264603e-05,
+      "loss": 1.2669,
+      "step": 9780
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.729444155071062e-05,
+      "loss": 1.2731,
+      "step": 9790
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7291465138775206e-05,
+      "loss": 1.2397,
+      "step": 9800
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.72884887268398e-05,
+      "loss": 1.2514,
+      "step": 9810
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.728551231490439e-05,
+      "loss": 1.261,
+      "step": 9820
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7282535902968974e-05,
+      "loss": 1.2573,
+      "step": 9830
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.727955949103356e-05,
+      "loss": 1.2488,
+      "step": 9840
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.7276583079098154e-05,
+      "loss": 1.2482,
+      "step": 9850
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.727360666716274e-05,
+      "loss": 1.2518,
+      "step": 9860
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.727063025522733e-05,
+      "loss": 1.2574,
+      "step": 9870
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.7267653843291915e-05,
+      "loss": 1.2535,
+      "step": 9880
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.72646774313565e-05,
+      "loss": 1.2407,
+      "step": 9890
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.726170101942109e-05,
+      "loss": 1.2641,
+      "step": 9900
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.725872460748568e-05,
+      "loss": 1.2625,
+      "step": 9910
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.725574819555027e-05,
+      "loss": 1.2705,
+      "step": 9920
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.7252771783614856e-05,
+      "loss": 1.2598,
+      "step": 9930
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.724979537167944e-05,
+      "loss": 1.2495,
+      "step": 9940
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.724681895974403e-05,
+      "loss": 1.2514,
+      "step": 9950
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.7243842547808624e-05,
+      "loss": 1.2362,
+      "step": 9960
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.7240866135873204e-05,
+      "loss": 1.2595,
+      "step": 9970
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.72378897239378e-05,
+      "loss": 1.2564,
+      "step": 9980
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.7234913312002384e-05,
+      "loss": 1.2687,
+      "step": 9990
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.723193690006697e-05,
+      "loss": 1.2474,
+      "step": 10000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.722896048813156e-05,
+      "loss": 1.25,
+      "step": 10010
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.722598407619615e-05,
+      "loss": 1.2333,
+      "step": 10020
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.722300766426074e-05,
+      "loss": 1.2758,
+      "step": 10030
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.7220031252325326e-05,
+      "loss": 1.2446,
+      "step": 10040
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.721705484038991e-05,
+      "loss": 1.2416,
+      "step": 10050
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.7214078428454506e-05,
+      "loss": 1.245,
+      "step": 10060
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.7211102016519086e-05,
+      "loss": 1.2408,
+      "step": 10070
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.720812560458367e-05,
+      "loss": 1.2515,
+      "step": 10080
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.720514919264827e-05,
+      "loss": 1.242,
+      "step": 10090
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.7202172780712854e-05,
+      "loss": 1.264,
+      "step": 10100
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.719919636877744e-05,
+      "loss": 1.2542,
+      "step": 10110
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.719621995684203e-05,
+      "loss": 1.2416,
+      "step": 10120
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.719324354490662e-05,
+      "loss": 1.253,
+      "step": 10130
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.719026713297121e-05,
+      "loss": 1.2612,
+      "step": 10140
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.7187290721035795e-05,
+      "loss": 1.2544,
+      "step": 10150
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.718431430910038e-05,
+      "loss": 1.2511,
+      "step": 10160
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.718133789716497e-05,
+      "loss": 1.2673,
+      "step": 10170
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.7178361485229556e-05,
+      "loss": 1.256,
+      "step": 10180
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.717538507329414e-05,
+      "loss": 1.2706,
+      "step": 10190
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.7172408661358736e-05,
+      "loss": 1.2486,
+      "step": 10200
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.716943224942332e-05,
+      "loss": 1.2476,
+      "step": 10210
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.716645583748791e-05,
+      "loss": 1.2431,
+      "step": 10220
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.71634794255525e-05,
+      "loss": 1.267,
+      "step": 10230
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.716050301361709e-05,
+      "loss": 1.2573,
+      "step": 10240
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.715752660168168e-05,
+      "loss": 1.2459,
+      "step": 10250
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.7154550189746264e-05,
+      "loss": 1.2415,
+      "step": 10260
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.715157377781085e-05,
+      "loss": 1.2614,
+      "step": 10270
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.714859736587544e-05,
+      "loss": 1.2414,
+      "step": 10280
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.7145620953940025e-05,
+      "loss": 1.2451,
+      "step": 10290
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.714264454200462e-05,
+      "loss": 1.2462,
+      "step": 10300
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.7139668130069206e-05,
+      "loss": 1.262,
+      "step": 10310
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.713669171813379e-05,
+      "loss": 1.2543,
+      "step": 10320
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.713371530619838e-05,
+      "loss": 1.2452,
+      "step": 10330
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.713073889426297e-05,
+      "loss": 1.2509,
+      "step": 10340
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.712776248232756e-05,
+      "loss": 1.2443,
+      "step": 10350
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.712478607039215e-05,
+      "loss": 1.2502,
+      "step": 10360
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.7121809658456734e-05,
+      "loss": 1.2459,
+      "step": 10370
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.711883324652132e-05,
+      "loss": 1.2411,
+      "step": 10380
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.711585683458591e-05,
+      "loss": 1.2533,
+      "step": 10390
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.7112880422650495e-05,
+      "loss": 1.2383,
+      "step": 10400
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.710990401071509e-05,
+      "loss": 1.2574,
+      "step": 10410
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.7106927598779675e-05,
+      "loss": 1.2529,
+      "step": 10420
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.710395118684426e-05,
+      "loss": 1.2569,
+      "step": 10430
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.710097477490885e-05,
+      "loss": 1.2733,
+      "step": 10440
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.709799836297344e-05,
+      "loss": 1.2558,
+      "step": 10450
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.709502195103803e-05,
+      "loss": 1.2477,
+      "step": 10460
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.7092045539102616e-05,
+      "loss": 1.2487,
+      "step": 10470
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.70890691271672e-05,
+      "loss": 1.2505,
+      "step": 10480
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.708609271523179e-05,
+      "loss": 1.2619,
+      "step": 10490
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.708311630329638e-05,
+      "loss": 1.2463,
+      "step": 10500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.7080139891360964e-05,
+      "loss": 1.2534,
+      "step": 10510
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.707716347942556e-05,
+      "loss": 1.2687,
+      "step": 10520
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.7074187067490145e-05,
+      "loss": 1.2644,
+      "step": 10530
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.707121065555473e-05,
+      "loss": 1.2634,
+      "step": 10540
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.706823424361932e-05,
+      "loss": 1.2507,
+      "step": 10550
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.706525783168391e-05,
+      "loss": 1.2382,
+      "step": 10560
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.70622814197485e-05,
+      "loss": 1.2662,
+      "step": 10570
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.705930500781308e-05,
+      "loss": 1.2534,
+      "step": 10580
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.705632859587767e-05,
+      "loss": 1.2561,
+      "step": 10590
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.705335218394226e-05,
+      "loss": 1.2401,
+      "step": 10600
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.7050375772006847e-05,
+      "loss": 1.2461,
+      "step": 10610
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.7047399360071433e-05,
+      "loss": 1.2454,
+      "step": 10620
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.704442294813603e-05,
+      "loss": 1.2449,
+      "step": 10630
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.7041446536200614e-05,
+      "loss": 1.2401,
+      "step": 10640
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.70384701242652e-05,
+      "loss": 1.2365,
+      "step": 10650
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.703549371232979e-05,
+      "loss": 1.2637,
+      "step": 10660
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.703251730039438e-05,
+      "loss": 1.2526,
+      "step": 10670
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.702954088845896e-05,
+      "loss": 1.2501,
+      "step": 10680
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.7026564476523555e-05,
+      "loss": 1.2432,
+      "step": 10690
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.702358806458814e-05,
+      "loss": 1.2649,
+      "step": 10700
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.702061165265273e-05,
+      "loss": 1.2698,
+      "step": 10710
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.7017635240717316e-05,
+      "loss": 1.2648,
+      "step": 10720
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.701465882878191e-05,
+      "loss": 1.2587,
+      "step": 10730
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.7011682416846497e-05,
+      "loss": 1.2565,
+      "step": 10740
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.7008706004911083e-05,
+      "loss": 1.2418,
+      "step": 10750
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.700572959297567e-05,
+      "loss": 1.2241,
+      "step": 10760
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.7002753181040264e-05,
+      "loss": 1.2446,
+      "step": 10770
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.699977676910485e-05,
+      "loss": 1.2379,
+      "step": 10780
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.699680035716943e-05,
+      "loss": 1.2459,
+      "step": 10790
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.6993823945234025e-05,
+      "loss": 1.2534,
+      "step": 10800
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.699084753329861e-05,
+      "loss": 1.2536,
+      "step": 10810
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.69878711213632e-05,
+      "loss": 1.2327,
+      "step": 10820
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.6984894709427785e-05,
+      "loss": 1.24,
+      "step": 10830
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.698191829749238e-05,
+      "loss": 1.242,
+      "step": 10840
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.6978941885556966e-05,
+      "loss": 1.2355,
+      "step": 10850
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.697596547362155e-05,
+      "loss": 1.2488,
+      "step": 10860
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.697298906168614e-05,
+      "loss": 1.2404,
+      "step": 10870
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.6970012649750733e-05,
+      "loss": 1.2336,
+      "step": 10880
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.6967036237815314e-05,
+      "loss": 1.2524,
+      "step": 10890
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.69640598258799e-05,
+      "loss": 1.2386,
+      "step": 10900
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.6961083413944494e-05,
+      "loss": 1.2455,
+      "step": 10910
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.695810700200908e-05,
+      "loss": 1.246,
+      "step": 10920
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.695513059007367e-05,
+      "loss": 1.2494,
+      "step": 10930
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.6952154178138255e-05,
+      "loss": 1.2576,
+      "step": 10940
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.694917776620285e-05,
+      "loss": 1.2712,
+      "step": 10950
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.6946201354267435e-05,
+      "loss": 1.2517,
+      "step": 10960
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.694322494233202e-05,
+      "loss": 1.2471,
+      "step": 10970
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.694024853039661e-05,
+      "loss": 1.2579,
+      "step": 10980
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.6937272118461196e-05,
+      "loss": 1.2556,
+      "step": 10990
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.693429570652578e-05,
+      "loss": 1.2626,
+      "step": 11000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.693131929459037e-05,
+      "loss": 1.2578,
+      "step": 11010
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6928342882654964e-05,
+      "loss": 1.2545,
+      "step": 11020
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.692536647071955e-05,
+      "loss": 1.2467,
+      "step": 11030
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.692239005878414e-05,
+      "loss": 1.2498,
+      "step": 11040
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6919413646848724e-05,
+      "loss": 1.2575,
+      "step": 11050
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.691643723491332e-05,
+      "loss": 1.2624,
+      "step": 11060
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6913460822977905e-05,
+      "loss": 1.2586,
+      "step": 11070
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.691048441104249e-05,
+      "loss": 1.2376,
+      "step": 11080
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.690750799910708e-05,
+      "loss": 1.2565,
+      "step": 11090
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6904531587171666e-05,
+      "loss": 1.2547,
+      "step": 11100
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.690155517523625e-05,
+      "loss": 1.2527,
+      "step": 11110
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6898578763300846e-05,
+      "loss": 1.2605,
+      "step": 11120
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.689560235136543e-05,
+      "loss": 1.2571,
+      "step": 11130
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.689262593943002e-05,
+      "loss": 1.2611,
+      "step": 11140
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.688964952749461e-05,
+      "loss": 1.236,
+      "step": 11150
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.68866731155592e-05,
+      "loss": 1.2416,
+      "step": 11160
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.688369670362379e-05,
+      "loss": 1.2523,
+      "step": 11170
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.6880720291688374e-05,
+      "loss": 1.2642,
+      "step": 11180
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.687774387975296e-05,
+      "loss": 1.2432,
+      "step": 11190
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.687476746781755e-05,
+      "loss": 1.245,
+      "step": 11200
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.6871791055882135e-05,
+      "loss": 1.2443,
+      "step": 11210
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.686881464394672e-05,
+      "loss": 1.2489,
+      "step": 11220
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.6865838232011316e-05,
+      "loss": 1.2466,
+      "step": 11230
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.68628618200759e-05,
+      "loss": 1.2258,
+      "step": 11240
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.685988540814049e-05,
+      "loss": 1.2391,
+      "step": 11250
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.6856908996205076e-05,
+      "loss": 1.2451,
+      "step": 11260
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.685393258426967e-05,
+      "loss": 1.2452,
+      "step": 11270
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.685095617233426e-05,
+      "loss": 1.2376,
+      "step": 11280
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.6847979760398844e-05,
+      "loss": 1.2376,
+      "step": 11290
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.684500334846343e-05,
+      "loss": 1.2461,
+      "step": 11300
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.684202693652802e-05,
+      "loss": 1.232,
+      "step": 11310
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.6839050524592604e-05,
+      "loss": 1.2349,
+      "step": 11320
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.683607411265719e-05,
+      "loss": 1.2367,
+      "step": 11330
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.6833097700721785e-05,
+      "loss": 1.2385,
+      "step": 11340
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.683012128878637e-05,
+      "loss": 1.2354,
+      "step": 11350
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.682714487685096e-05,
+      "loss": 1.2343,
+      "step": 11360
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.6824168464915546e-05,
+      "loss": 1.2398,
+      "step": 11370
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.682119205298014e-05,
+      "loss": 1.2346,
+      "step": 11380
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.6818215641044726e-05,
+      "loss": 1.2526,
+      "step": 11390
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.6815239229109306e-05,
+      "loss": 1.2589,
+      "step": 11400
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.68122628171739e-05,
+      "loss": 1.2495,
+      "step": 11410
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.680928640523849e-05,
+      "loss": 1.2361,
+      "step": 11420
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.6806309993303074e-05,
+      "loss": 1.2428,
+      "step": 11430
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.680333358136766e-05,
+      "loss": 1.2177,
+      "step": 11440
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.6800357169432254e-05,
+      "loss": 1.2531,
+      "step": 11450
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.679738075749684e-05,
+      "loss": 1.2292,
+      "step": 11460
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.679440434556143e-05,
+      "loss": 1.2267,
+      "step": 11470
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.6791427933626015e-05,
+      "loss": 1.2392,
+      "step": 11480
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.678845152169061e-05,
+      "loss": 1.2413,
+      "step": 11490
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.678547510975519e-05,
+      "loss": 1.2434,
+      "step": 11500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.678249869781978e-05,
+      "loss": 1.2287,
+      "step": 11510
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.677952228588437e-05,
+      "loss": 1.2341,
+      "step": 11520
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.6776545873948956e-05,
+      "loss": 1.2381,
+      "step": 11530
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.677356946201354e-05,
+      "loss": 1.2676,
+      "step": 11540
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.677059305007814e-05,
+      "loss": 1.2371,
+      "step": 11550
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.6767616638142724e-05,
+      "loss": 1.2447,
+      "step": 11560
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.676464022620731e-05,
+      "loss": 1.2219,
+      "step": 11570
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.67616638142719e-05,
+      "loss": 1.2359,
+      "step": 11580
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.675868740233649e-05,
+      "loss": 1.2286,
+      "step": 11590
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.675571099040108e-05,
+      "loss": 1.2417,
+      "step": 11600
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.675273457846566e-05,
+      "loss": 1.2229,
+      "step": 11610
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.674975816653025e-05,
+      "loss": 1.239,
+      "step": 11620
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.674678175459484e-05,
+      "loss": 1.2435,
+      "step": 11630
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.6743805342659426e-05,
+      "loss": 1.229,
+      "step": 11640
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.674082893072401e-05,
+      "loss": 1.2191,
+      "step": 11650
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.6737852518788606e-05,
+      "loss": 1.2414,
+      "step": 11660
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.673487610685319e-05,
+      "loss": 1.2254,
+      "step": 11670
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.673189969491778e-05,
+      "loss": 1.2199,
+      "step": 11680
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.672892328298237e-05,
+      "loss": 1.231,
+      "step": 11690
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.672594687104696e-05,
+      "loss": 1.2246,
+      "step": 11700
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.672297045911154e-05,
+      "loss": 1.2413,
+      "step": 11710
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.671999404717613e-05,
+      "loss": 1.2359,
+      "step": 11720
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.671701763524072e-05,
+      "loss": 1.2369,
+      "step": 11730
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.671404122330531e-05,
+      "loss": 1.2442,
+      "step": 11740
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6711064811369895e-05,
+      "loss": 1.2302,
+      "step": 11750
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.670808839943448e-05,
+      "loss": 1.2309,
+      "step": 11760
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6705111987499076e-05,
+      "loss": 1.2334,
+      "step": 11770
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.670213557556366e-05,
+      "loss": 1.2213,
+      "step": 11780
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.669915916362825e-05,
+      "loss": 1.2364,
+      "step": 11790
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6696182751692837e-05,
+      "loss": 1.2365,
+      "step": 11800
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6693206339757423e-05,
+      "loss": 1.2535,
+      "step": 11810
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.669022992782201e-05,
+      "loss": 1.2428,
+      "step": 11820
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.66872535158866e-05,
+      "loss": 1.2411,
+      "step": 11830
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.668427710395119e-05,
+      "loss": 1.2534,
+      "step": 11840
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.668130069201578e-05,
+      "loss": 1.2295,
+      "step": 11850
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6678324280080365e-05,
+      "loss": 1.2257,
+      "step": 11860
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.667534786814495e-05,
+      "loss": 1.2379,
+      "step": 11870
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6672371456209545e-05,
+      "loss": 1.225,
+      "step": 11880
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.666939504427413e-05,
+      "loss": 1.2338,
+      "step": 11890
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.666641863233872e-05,
+      "loss": 1.232,
+      "step": 11900
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6663442220403306e-05,
+      "loss": 1.2355,
+      "step": 11910
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.666046580846789e-05,
+      "loss": 1.2336,
+      "step": 11920
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.665748939653248e-05,
+      "loss": 1.2403,
+      "step": 11930
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6654512984597074e-05,
+      "loss": 1.2178,
+      "step": 11940
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.665153657266166e-05,
+      "loss": 1.235,
+      "step": 11950
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.664856016072625e-05,
+      "loss": 1.2366,
+      "step": 11960
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6645583748790834e-05,
+      "loss": 1.2258,
+      "step": 11970
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.664260733685543e-05,
+      "loss": 1.2286,
+      "step": 11980
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6639630924920015e-05,
+      "loss": 1.2458,
+      "step": 11990
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.66366545129846e-05,
+      "loss": 1.2447,
+      "step": 12000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.663367810104919e-05,
+      "loss": 1.2407,
+      "step": 12010
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6630701689113775e-05,
+      "loss": 1.2314,
+      "step": 12020
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.662772527717836e-05,
+      "loss": 1.2324,
+      "step": 12030
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.662474886524295e-05,
+      "loss": 1.228,
+      "step": 12040
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.662177245330754e-05,
+      "loss": 1.2236,
+      "step": 12050
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.661879604137213e-05,
+      "loss": 1.2451,
+      "step": 12060
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.661581962943672e-05,
+      "loss": 1.2387,
+      "step": 12070
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6612843217501304e-05,
+      "loss": 1.2152,
+      "step": 12080
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.66098668055659e-05,
+      "loss": 1.2251,
+      "step": 12090
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6606890393630484e-05,
+      "loss": 1.2259,
+      "step": 12100
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.660391398169507e-05,
+      "loss": 1.225,
+      "step": 12110
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.660093756975966e-05,
+      "loss": 1.2268,
+      "step": 12120
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6597961157824245e-05,
+      "loss": 1.2385,
+      "step": 12130
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.659498474588883e-05,
+      "loss": 1.2406,
+      "step": 12140
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.659200833395342e-05,
+      "loss": 1.2266,
+      "step": 12150
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.658903192201801e-05,
+      "loss": 1.2333,
+      "step": 12160
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.65860555100826e-05,
+      "loss": 1.2511,
+      "step": 12170
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6583079098147186e-05,
+      "loss": 1.2364,
+      "step": 12180
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.658010268621177e-05,
+      "loss": 1.2442,
+      "step": 12190
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.657712627427637e-05,
+      "loss": 1.2492,
+      "step": 12200
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6574149862340954e-05,
+      "loss": 1.2322,
+      "step": 12210
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6571173450405534e-05,
+      "loss": 1.2385,
+      "step": 12220
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.656819703847013e-05,
+      "loss": 1.2306,
+      "step": 12230
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6565220626534714e-05,
+      "loss": 1.2357,
+      "step": 12240
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.65622442145993e-05,
+      "loss": 1.2364,
+      "step": 12250
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.655926780266389e-05,
+      "loss": 1.2252,
+      "step": 12260
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.655629139072848e-05,
+      "loss": 1.2346,
+      "step": 12270
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.655331497879307e-05,
+      "loss": 1.2358,
+      "step": 12280
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6550338566857656e-05,
+      "loss": 1.2522,
+      "step": 12290
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.654736215492224e-05,
+      "loss": 1.2272,
+      "step": 12300
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6544385742986836e-05,
+      "loss": 1.2382,
+      "step": 12310
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6541409331051416e-05,
+      "loss": 1.2342,
+      "step": 12320
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.653843291911601e-05,
+      "loss": 1.24,
+      "step": 12330
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.65354565071806e-05,
+      "loss": 1.2348,
+      "step": 12340
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6532480095245184e-05,
+      "loss": 1.238,
+      "step": 12350
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.652950368330977e-05,
+      "loss": 1.2574,
+      "step": 12360
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6526527271374364e-05,
+      "loss": 1.233,
+      "step": 12370
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.652355085943895e-05,
+      "loss": 1.2441,
+      "step": 12380
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.652057444750354e-05,
+      "loss": 1.2206,
+      "step": 12390
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6517598035568125e-05,
+      "loss": 1.2248,
+      "step": 12400
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.651462162363272e-05,
+      "loss": 1.2375,
+      "step": 12410
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.65116452116973e-05,
+      "loss": 1.2331,
+      "step": 12420
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6508668799761886e-05,
+      "loss": 1.2266,
+      "step": 12430
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.650569238782648e-05,
+      "loss": 1.2379,
+      "step": 12440
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6502715975891066e-05,
+      "loss": 1.2227,
+      "step": 12450
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.649973956395565e-05,
+      "loss": 1.222,
+      "step": 12460
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.649676315202024e-05,
+      "loss": 1.2242,
+      "step": 12470
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6493786740084834e-05,
+      "loss": 1.2301,
+      "step": 12480
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.649081032814942e-05,
+      "loss": 1.2246,
+      "step": 12490
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.648783391621401e-05,
+      "loss": 1.232,
+      "step": 12500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6484857504278595e-05,
+      "loss": 1.2185,
+      "step": 12510
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.648188109234319e-05,
+      "loss": 1.2182,
+      "step": 12520
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.647890468040777e-05,
+      "loss": 1.2057,
+      "step": 12530
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6475928268472355e-05,
+      "loss": 1.2351,
+      "step": 12540
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.647295185653695e-05,
+      "loss": 1.2129,
+      "step": 12550
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6469975444601536e-05,
+      "loss": 1.2195,
+      "step": 12560
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.646699903266612e-05,
+      "loss": 1.2167,
+      "step": 12570
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.646402262073071e-05,
+      "loss": 1.2205,
+      "step": 12580
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.64610462087953e-05,
+      "loss": 1.222,
+      "step": 12590
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.645806979685989e-05,
+      "loss": 1.2334,
+      "step": 12600
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.645509338492448e-05,
+      "loss": 1.2358,
+      "step": 12610
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6452116972989064e-05,
+      "loss": 1.2172,
+      "step": 12620
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.644914056105365e-05,
+      "loss": 1.2186,
+      "step": 12630
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.644616414911824e-05,
+      "loss": 1.2305,
+      "step": 12640
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6443187737182825e-05,
+      "loss": 1.2248,
+      "step": 12650
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.644021132524742e-05,
+      "loss": 1.2135,
+      "step": 12660
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6437234913312005e-05,
+      "loss": 1.2152,
+      "step": 12670
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.643425850137659e-05,
+      "loss": 1.2125,
+      "step": 12680
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.643128208944118e-05,
+      "loss": 1.2235,
+      "step": 12690
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.642830567750577e-05,
+      "loss": 1.2361,
+      "step": 12700
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.642532926557036e-05,
+      "loss": 1.2135,
+      "step": 12710
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.6422352853634946e-05,
+      "loss": 1.2292,
+      "step": 12720
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.6419376441699533e-05,
+      "loss": 1.234,
+      "step": 12730
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.641640002976412e-05,
+      "loss": 1.2145,
+      "step": 12740
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.641342361782871e-05,
+      "loss": 1.2284,
+      "step": 12750
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.64104472058933e-05,
+      "loss": 1.2216,
+      "step": 12760
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.640747079395789e-05,
+      "loss": 1.228,
+      "step": 12770
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6404494382022475e-05,
+      "loss": 1.2217,
+      "step": 12780
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.640151797008706e-05,
+      "loss": 1.2256,
+      "step": 12790
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6398541558151655e-05,
+      "loss": 1.2262,
+      "step": 12800
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.639556514621624e-05,
+      "loss": 1.2169,
+      "step": 12810
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.639258873428083e-05,
+      "loss": 1.222,
+      "step": 12820
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6389612322345416e-05,
+      "loss": 1.219,
+      "step": 12830
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.638663591041e-05,
+      "loss": 1.2281,
+      "step": 12840
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.638365949847459e-05,
+      "loss": 1.2131,
+      "step": 12850
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.638068308653918e-05,
+      "loss": 1.2179,
+      "step": 12860
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.637770667460377e-05,
+      "loss": 1.2305,
+      "step": 12870
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.637473026266836e-05,
+      "loss": 1.2263,
+      "step": 12880
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.6371753850732944e-05,
+      "loss": 1.217,
+      "step": 12890
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.636877743879753e-05,
+      "loss": 1.2278,
+      "step": 12900
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.6365801026862125e-05,
+      "loss": 1.2321,
+      "step": 12910
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.636282461492671e-05,
+      "loss": 1.2163,
+      "step": 12920
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.63598482029913e-05,
+      "loss": 1.2415,
+      "step": 12930
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.6356871791055885e-05,
+      "loss": 1.225,
+      "step": 12940
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.635389537912047e-05,
+      "loss": 1.2145,
+      "step": 12950
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.635091896718506e-05,
+      "loss": 1.2261,
+      "step": 12960
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.6347942555249646e-05,
+      "loss": 1.2114,
+      "step": 12970
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.634496614331424e-05,
+      "loss": 1.2265,
+      "step": 12980
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.634198973137883e-05,
+      "loss": 1.2239,
+      "step": 12990
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.6339013319443414e-05,
+      "loss": 1.2336,
+      "step": 13000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.6336036907508e-05,
+      "loss": 1.211,
+      "step": 13010
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.6333060495572594e-05,
+      "loss": 1.2109,
+      "step": 13020
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.633008408363718e-05,
+      "loss": 1.2242,
+      "step": 13030
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.632710767170177e-05,
+      "loss": 1.2168,
+      "step": 13040
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.6324131259766355e-05,
+      "loss": 1.2141,
+      "step": 13050
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.632115484783094e-05,
+      "loss": 1.224,
+      "step": 13060
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.631817843589553e-05,
+      "loss": 1.2276,
+      "step": 13070
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.6315202023960116e-05,
+      "loss": 1.2124,
+      "step": 13080
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.631222561202471e-05,
+      "loss": 1.2196,
+      "step": 13090
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.6309249200089296e-05,
+      "loss": 1.219,
+      "step": 13100
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.630627278815388e-05,
+      "loss": 1.2121,
+      "step": 13110
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.630329637621847e-05,
+      "loss": 1.2178,
+      "step": 13120
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6300319964283064e-05,
+      "loss": 1.2158,
+      "step": 13130
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6297343552347644e-05,
+      "loss": 1.2082,
+      "step": 13140
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.629436714041224e-05,
+      "loss": 1.2043,
+      "step": 13150
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6291390728476824e-05,
+      "loss": 1.1986,
+      "step": 13160
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.628841431654141e-05,
+      "loss": 1.223,
+      "step": 13170
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.6285437904606e-05,
+      "loss": 1.2173,
+      "step": 13180
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.628246149267059e-05,
+      "loss": 1.2209,
+      "step": 13190
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.627948508073518e-05,
+      "loss": 1.2258,
+      "step": 13200
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.6276508668799766e-05,
+      "loss": 1.2111,
+      "step": 13210
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.627353225686435e-05,
+      "loss": 1.213,
+      "step": 13220
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.6270555844928946e-05,
+      "loss": 1.2004,
+      "step": 13230
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.6267579432993526e-05,
+      "loss": 1.2165,
+      "step": 13240
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.626460302105811e-05,
+      "loss": 1.2248,
+      "step": 13250
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.626162660912271e-05,
+      "loss": 1.2226,
+      "step": 13260
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.6258650197187294e-05,
+      "loss": 1.2109,
+      "step": 13270
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.625567378525188e-05,
+      "loss": 1.2355,
+      "step": 13280
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.625269737331647e-05,
+      "loss": 1.2139,
+      "step": 13290
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.624972096138106e-05,
+      "loss": 1.2093,
+      "step": 13300
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.624674454944565e-05,
+      "loss": 1.2243,
+      "step": 13310
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.6243768137510235e-05,
+      "loss": 1.2246,
+      "step": 13320
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.624079172557482e-05,
+      "loss": 1.2117,
+      "step": 13330
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.6237815313639416e-05,
+      "loss": 1.2253,
+      "step": 13340
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.6234838901703996e-05,
+      "loss": 1.2243,
+      "step": 13350
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.623186248976858e-05,
+      "loss": 1.2322,
+      "step": 13360
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.6228886077833176e-05,
+      "loss": 1.2161,
+      "step": 13370
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.622590966589776e-05,
+      "loss": 1.2116,
+      "step": 13380
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.622293325396235e-05,
+      "loss": 1.2142,
+      "step": 13390
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.621995684202694e-05,
+      "loss": 1.205,
+      "step": 13400
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.621698043009153e-05,
+      "loss": 1.2235,
+      "step": 13410
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.621400401815612e-05,
+      "loss": 1.2307,
+      "step": 13420
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.6211027606220704e-05,
+      "loss": 1.2198,
+      "step": 13430
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.620805119428529e-05,
+      "loss": 1.2243,
+      "step": 13440
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.620507478234988e-05,
+      "loss": 1.2088,
+      "step": 13450
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.6202098370414465e-05,
+      "loss": 1.2313,
+      "step": 13460
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.619912195847906e-05,
+      "loss": 1.2202,
+      "step": 13470
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.6196145546543646e-05,
+      "loss": 1.2126,
+      "step": 13480
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.619316913460823e-05,
+      "loss": 1.2121,
+      "step": 13490
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.619019272267282e-05,
+      "loss": 1.2244,
+      "step": 13500
+    },
+    {
+      "epoch": 3.0,
+      "eval_cer": 4.7756410256410255,
+      "eval_loss": 1.1598864793777466,
+      "eval_runtime": 5.7385,
+      "eval_samples_per_second": 1.743,
+      "eval_steps_per_second": 0.174,
+      "eval_wer": 1.0129870129870129,
+      "step": 13509
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.6187216310737406e-05,
+      "loss": 1.2271,
+      "step": 13510
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.6184239898802e-05,
+      "loss": 1.2057,
+      "step": 13520
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.618126348686659e-05,
+      "loss": 1.2122,
+      "step": 13530
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.6178287074931174e-05,
+      "loss": 1.1948,
+      "step": 13540
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.617531066299576e-05,
+      "loss": 1.209,
+      "step": 13550
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.617233425106035e-05,
+      "loss": 1.2247,
+      "step": 13560
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.6169357839124935e-05,
+      "loss": 1.2267,
+      "step": 13570
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.616638142718953e-05,
+      "loss": 1.2131,
+      "step": 13580
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.6163405015254115e-05,
+      "loss": 1.2173,
+      "step": 13590
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.61604286033187e-05,
+      "loss": 1.1959,
+      "step": 13600
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.615745219138329e-05,
+      "loss": 1.2225,
+      "step": 13610
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.615447577944788e-05,
+      "loss": 1.2228,
+      "step": 13620
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.615149936751247e-05,
+      "loss": 1.2146,
+      "step": 13630
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.6148522955577056e-05,
+      "loss": 1.2201,
+      "step": 13640
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.614554654364164e-05,
+      "loss": 1.2042,
+      "step": 13650
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.614257013170623e-05,
+      "loss": 1.2273,
+      "step": 13660
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.613959371977082e-05,
+      "loss": 1.2099,
+      "step": 13670
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.6136617307835404e-05,
+      "loss": 1.1863,
+      "step": 13680
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.61336408959e-05,
+      "loss": 1.2294,
+      "step": 13690
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.6130664483964585e-05,
+      "loss": 1.222,
+      "step": 13700
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.612768807202917e-05,
+      "loss": 1.2148,
+      "step": 13710
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.612471166009376e-05,
+      "loss": 1.2259,
+      "step": 13720
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.612173524815835e-05,
+      "loss": 1.2151,
+      "step": 13730
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.611875883622294e-05,
+      "loss": 1.2009,
+      "step": 13740
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.6115782424287526e-05,
+      "loss": 1.2102,
+      "step": 13750
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.611280601235211e-05,
+      "loss": 1.2143,
+      "step": 13760
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.61098296004167e-05,
+      "loss": 1.2165,
+      "step": 13770
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.6106853188481287e-05,
+      "loss": 1.2236,
+      "step": 13780
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.6103876776545873e-05,
+      "loss": 1.2136,
+      "step": 13790
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.610090036461047e-05,
+      "loss": 1.2222,
+      "step": 13800
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.6097923952675054e-05,
+      "loss": 1.2171,
+      "step": 13810
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.609494754073964e-05,
+      "loss": 1.2057,
+      "step": 13820
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.609197112880423e-05,
+      "loss": 1.208,
+      "step": 13830
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.608899471686882e-05,
+      "loss": 1.2084,
+      "step": 13840
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.608601830493341e-05,
+      "loss": 1.2025,
+      "step": 13850
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.6083041892997995e-05,
+      "loss": 1.2069,
+      "step": 13860
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.608006548106258e-05,
+      "loss": 1.2022,
+      "step": 13870
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.607708906912717e-05,
+      "loss": 1.2082,
+      "step": 13880
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.6074112657191756e-05,
+      "loss": 1.2057,
+      "step": 13890
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.607113624525635e-05,
+      "loss": 1.202,
+      "step": 13900
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.6068159833320937e-05,
+      "loss": 1.2071,
+      "step": 13910
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.6065183421385523e-05,
+      "loss": 1.2151,
+      "step": 13920
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.606220700945011e-05,
+      "loss": 1.2176,
+      "step": 13930
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.60592305975147e-05,
+      "loss": 1.2137,
+      "step": 13940
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.605625418557929e-05,
+      "loss": 1.2228,
+      "step": 13950
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.605327777364387e-05,
+      "loss": 1.2112,
+      "step": 13960
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.6050301361708465e-05,
+      "loss": 1.2137,
+      "step": 13970
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.604732494977305e-05,
+      "loss": 1.2075,
+      "step": 13980
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.604434853783764e-05,
+      "loss": 1.2172,
+      "step": 13990
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.6041372125902225e-05,
+      "loss": 1.2079,
+      "step": 14000
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.603839571396682e-05,
+      "loss": 1.2106,
+      "step": 14010
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.6035419302031406e-05,
+      "loss": 1.2007,
+      "step": 14020
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.603244289009599e-05,
+      "loss": 1.2022,
+      "step": 14030
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.602946647816058e-05,
+      "loss": 1.2003,
+      "step": 14040
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.6026490066225173e-05,
+      "loss": 1.211,
+      "step": 14050
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.6023513654289754e-05,
+      "loss": 1.2116,
+      "step": 14060
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.602053724235434e-05,
+      "loss": 1.2088,
+      "step": 14070
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.6017560830418934e-05,
+      "loss": 1.2119,
+      "step": 14080
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.601458441848352e-05,
+      "loss": 1.2036,
+      "step": 14090
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.601160800654811e-05,
+      "loss": 1.2048,
+      "step": 14100
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.6008631594612695e-05,
+      "loss": 1.2111,
+      "step": 14110
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.600565518267729e-05,
+      "loss": 1.2005,
+      "step": 14120
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.6002678770741875e-05,
+      "loss": 1.2072,
+      "step": 14130
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.599970235880646e-05,
+      "loss": 1.2217,
+      "step": 14140
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.599672594687105e-05,
+      "loss": 1.2169,
+      "step": 14150
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.5993749534935636e-05,
+      "loss": 1.2045,
+      "step": 14160
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.599077312300022e-05,
+      "loss": 1.2103,
+      "step": 14170
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.598779671106481e-05,
+      "loss": 1.1995,
+      "step": 14180
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.5984820299129404e-05,
+      "loss": 1.2288,
+      "step": 14190
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.598184388719399e-05,
+      "loss": 1.2078,
+      "step": 14200
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.597886747525858e-05,
+      "loss": 1.2169,
+      "step": 14210
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.5975891063323164e-05,
+      "loss": 1.2097,
+      "step": 14220
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.597291465138776e-05,
+      "loss": 1.2151,
+      "step": 14230
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.5969938239452345e-05,
+      "loss": 1.2018,
+      "step": 14240
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.596696182751693e-05,
+      "loss": 1.1763,
+      "step": 14250
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.596398541558152e-05,
+      "loss": 1.2096,
+      "step": 14260
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.5961009003646106e-05,
+      "loss": 1.2086,
+      "step": 14270
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.595803259171069e-05,
+      "loss": 1.222,
+      "step": 14280
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.5955056179775286e-05,
+      "loss": 1.217,
+      "step": 14290
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.595207976783987e-05,
+      "loss": 1.2052,
+      "step": 14300
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.594910335590446e-05,
+      "loss": 1.209,
+      "step": 14310
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.594612694396905e-05,
+      "loss": 1.2168,
+      "step": 14320
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.594315053203364e-05,
+      "loss": 1.1888,
+      "step": 14330
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.594017412009823e-05,
+      "loss": 1.22,
+      "step": 14340
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.5937197708162814e-05,
+      "loss": 1.2319,
+      "step": 14350
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.59342212962274e-05,
+      "loss": 1.1983,
+      "step": 14360
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.593124488429199e-05,
+      "loss": 1.2098,
+      "step": 14370
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.5928268472356575e-05,
+      "loss": 1.2253,
+      "step": 14380
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.592529206042116e-05,
+      "loss": 1.2082,
+      "step": 14390
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.5922315648485756e-05,
+      "loss": 1.2144,
+      "step": 14400
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.591933923655034e-05,
+      "loss": 1.2106,
+      "step": 14410
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.591636282461493e-05,
+      "loss": 1.1985,
+      "step": 14420
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.5913386412679516e-05,
+      "loss": 1.2083,
+      "step": 14430
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.591041000074411e-05,
+      "loss": 1.2085,
+      "step": 14440
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.59074335888087e-05,
+      "loss": 1.2069,
+      "step": 14450
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.5904457176873284e-05,
+      "loss": 1.2176,
+      "step": 14460
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.590148076493787e-05,
+      "loss": 1.1939,
+      "step": 14470
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.589850435300246e-05,
+      "loss": 1.1968,
+      "step": 14480
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.5895527941067044e-05,
+      "loss": 1.1975,
+      "step": 14490
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.589255152913163e-05,
+      "loss": 1.21,
+      "step": 14500
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.5889575117196225e-05,
+      "loss": 1.2095,
+      "step": 14510
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.588659870526081e-05,
+      "loss": 1.2162,
+      "step": 14520
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.58836222933254e-05,
+      "loss": 1.2086,
+      "step": 14530
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.5880645881389986e-05,
+      "loss": 1.2144,
+      "step": 14540
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.587766946945458e-05,
+      "loss": 1.2114,
+      "step": 14550
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.5874693057519166e-05,
+      "loss": 1.2134,
+      "step": 14560
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.5871716645583746e-05,
+      "loss": 1.201,
+      "step": 14570
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.586874023364834e-05,
+      "loss": 1.2046,
+      "step": 14580
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.586576382171293e-05,
+      "loss": 1.2114,
+      "step": 14590
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.5862787409777514e-05,
+      "loss": 1.1977,
+      "step": 14600
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.58598109978421e-05,
+      "loss": 1.2025,
+      "step": 14610
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.5856834585906694e-05,
+      "loss": 1.21,
+      "step": 14620
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.585385817397128e-05,
+      "loss": 1.2088,
+      "step": 14630
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.585088176203587e-05,
+      "loss": 1.2156,
+      "step": 14640
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.5847905350100455e-05,
+      "loss": 1.1987,
+      "step": 14650
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.584492893816505e-05,
+      "loss": 1.201,
+      "step": 14660
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.5841952526229636e-05,
+      "loss": 1.2093,
+      "step": 14670
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.583897611429422e-05,
+      "loss": 1.2011,
+      "step": 14680
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.583599970235881e-05,
+      "loss": 1.1983,
+      "step": 14690
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.5833023290423396e-05,
+      "loss": 1.1956,
+      "step": 14700
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.583004687848798e-05,
+      "loss": 1.1993,
+      "step": 14710
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.582707046655258e-05,
+      "loss": 1.2105,
+      "step": 14720
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.5824094054617164e-05,
+      "loss": 1.2131,
+      "step": 14730
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.582111764268175e-05,
+      "loss": 1.2085,
+      "step": 14740
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.581814123074634e-05,
+      "loss": 1.209,
+      "step": 14750
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.581516481881093e-05,
+      "loss": 1.2215,
+      "step": 14760
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.581218840687552e-05,
+      "loss": 1.2089,
+      "step": 14770
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.58092119949401e-05,
+      "loss": 1.2229,
+      "step": 14780
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.580623558300469e-05,
+      "loss": 1.2025,
+      "step": 14790
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.580325917106928e-05,
+      "loss": 1.1991,
+      "step": 14800
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.5800282759133866e-05,
+      "loss": 1.2138,
+      "step": 14810
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.579730634719845e-05,
+      "loss": 1.2137,
+      "step": 14820
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.5794329935263046e-05,
+      "loss": 1.2151,
+      "step": 14830
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.579135352332763e-05,
+      "loss": 1.2109,
+      "step": 14840
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.578837711139222e-05,
+      "loss": 1.2012,
+      "step": 14850
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.578540069945681e-05,
+      "loss": 1.2027,
+      "step": 14860
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.57824242875214e-05,
+      "loss": 1.1981,
+      "step": 14870
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.577944787558598e-05,
+      "loss": 1.2079,
+      "step": 14880
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.577647146365057e-05,
+      "loss": 1.1946,
+      "step": 14890
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.577349505171516e-05,
+      "loss": 1.2182,
+      "step": 14900
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.577051863977975e-05,
+      "loss": 1.2133,
+      "step": 14910
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.5767542227844335e-05,
+      "loss": 1.1944,
+      "step": 14920
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.576456581590892e-05,
+      "loss": 1.1903,
+      "step": 14930
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.5761589403973516e-05,
+      "loss": 1.1949,
+      "step": 14940
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.57586129920381e-05,
+      "loss": 1.1931,
+      "step": 14950
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.575563658010269e-05,
+      "loss": 1.2279,
+      "step": 14960
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.5752660168167277e-05,
+      "loss": 1.218,
+      "step": 14970
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.5749683756231863e-05,
+      "loss": 1.1998,
+      "step": 14980
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.574670734429645e-05,
+      "loss": 1.1905,
+      "step": 14990
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.574373093236104e-05,
+      "loss": 1.1938,
+      "step": 15000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.574075452042563e-05,
+      "loss": 1.2028,
+      "step": 15010
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.573777810849022e-05,
+      "loss": 1.2004,
+      "step": 15020
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.5734801696554805e-05,
+      "loss": 1.211,
+      "step": 15030
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.573182528461939e-05,
+      "loss": 1.1959,
+      "step": 15040
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.5728848872683985e-05,
+      "loss": 1.1935,
+      "step": 15050
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.572587246074857e-05,
+      "loss": 1.1947,
+      "step": 15060
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.572289604881316e-05,
+      "loss": 1.1991,
+      "step": 15070
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.5719919636877746e-05,
+      "loss": 1.2058,
+      "step": 15080
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.571694322494233e-05,
+      "loss": 1.2008,
+      "step": 15090
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.571396681300692e-05,
+      "loss": 1.1774,
+      "step": 15100
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.5710990401071513e-05,
+      "loss": 1.1968,
+      "step": 15110
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.57080139891361e-05,
+      "loss": 1.2118,
+      "step": 15120
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.570503757720069e-05,
+      "loss": 1.1906,
+      "step": 15130
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.5702061165265274e-05,
+      "loss": 1.1935,
+      "step": 15140
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.569908475332987e-05,
+      "loss": 1.1967,
+      "step": 15150
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.5696108341394455e-05,
+      "loss": 1.1948,
+      "step": 15160
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.569313192945904e-05,
+      "loss": 1.202,
+      "step": 15170
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.569015551752363e-05,
+      "loss": 1.1985,
+      "step": 15180
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.5687179105588215e-05,
+      "loss": 1.2098,
+      "step": 15190
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.56842026936528e-05,
+      "loss": 1.198,
+      "step": 15200
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.568122628171739e-05,
+      "loss": 1.2046,
+      "step": 15210
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.567824986978198e-05,
+      "loss": 1.2095,
+      "step": 15220
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.567527345784657e-05,
+      "loss": 1.1965,
+      "step": 15230
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.567229704591116e-05,
+      "loss": 1.1886,
+      "step": 15240
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.5669320633975744e-05,
+      "loss": 1.1839,
+      "step": 15250
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.566634422204034e-05,
+      "loss": 1.1989,
+      "step": 15260
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.5663367810104924e-05,
+      "loss": 1.1975,
+      "step": 15270
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.566039139816951e-05,
+      "loss": 1.2085,
+      "step": 15280
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.56574149862341e-05,
+      "loss": 1.2091,
+      "step": 15290
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.5654438574298685e-05,
+      "loss": 1.1982,
+      "step": 15300
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.565146216236327e-05,
+      "loss": 1.202,
+      "step": 15310
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.564848575042786e-05,
+      "loss": 1.1854,
+      "step": 15320
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.564550933849245e-05,
+      "loss": 1.2018,
+      "step": 15330
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.564253292655704e-05,
+      "loss": 1.1962,
+      "step": 15340
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.5639556514621626e-05,
+      "loss": 1.1962,
+      "step": 15350
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.563658010268621e-05,
+      "loss": 1.1879,
+      "step": 15360
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.563360369075081e-05,
+      "loss": 1.1959,
+      "step": 15370
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.5630627278815394e-05,
+      "loss": 1.2094,
+      "step": 15380
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.5627650866879974e-05,
+      "loss": 1.1879,
+      "step": 15390
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.562467445494457e-05,
+      "loss": 1.201,
+      "step": 15400
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.5621698043009154e-05,
+      "loss": 1.2081,
+      "step": 15410
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.561872163107374e-05,
+      "loss": 1.1902,
+      "step": 15420
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.561574521913833e-05,
+      "loss": 1.1996,
+      "step": 15430
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.561276880720292e-05,
+      "loss": 1.1864,
+      "step": 15440
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.560979239526751e-05,
+      "loss": 1.198,
+      "step": 15450
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.5606815983332096e-05,
+      "loss": 1.2004,
+      "step": 15460
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.560383957139668e-05,
+      "loss": 1.211,
+      "step": 15470
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.5600863159461276e-05,
+      "loss": 1.1997,
+      "step": 15480
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.559788674752586e-05,
+      "loss": 1.193,
+      "step": 15490
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.559491033559045e-05,
+      "loss": 1.204,
+      "step": 15500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.559193392365504e-05,
+      "loss": 1.1936,
+      "step": 15510
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.5588957511719624e-05,
+      "loss": 1.1893,
+      "step": 15520
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.558598109978421e-05,
+      "loss": 1.2024,
+      "step": 15530
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.5583004687848804e-05,
+      "loss": 1.2036,
+      "step": 15540
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.558002827591339e-05,
+      "loss": 1.1928,
+      "step": 15550
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.557705186397798e-05,
+      "loss": 1.1936,
+      "step": 15560
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.5574075452042565e-05,
+      "loss": 1.2062,
+      "step": 15570
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.557109904010716e-05,
+      "loss": 1.2006,
+      "step": 15580
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.5568122628171746e-05,
+      "loss": 1.204,
+      "step": 15590
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.5565146216236326e-05,
+      "loss": 1.1917,
+      "step": 15600
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.556216980430092e-05,
+      "loss": 1.2094,
+      "step": 15610
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.5559193392365506e-05,
+      "loss": 1.197,
+      "step": 15620
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.555621698043009e-05,
+      "loss": 1.213,
+      "step": 15630
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.555324056849468e-05,
+      "loss": 1.1888,
+      "step": 15640
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.5550264156559274e-05,
+      "loss": 1.1919,
+      "step": 15650
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.554728774462386e-05,
+      "loss": 1.2013,
+      "step": 15660
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.554431133268845e-05,
+      "loss": 1.182,
+      "step": 15670
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.5541334920753034e-05,
+      "loss": 1.1977,
+      "step": 15680
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.553835850881763e-05,
+      "loss": 1.1892,
+      "step": 15690
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.553538209688221e-05,
+      "loss": 1.1752,
+      "step": 15700
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.5532405684946795e-05,
+      "loss": 1.1939,
+      "step": 15710
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.552942927301139e-05,
+      "loss": 1.1829,
+      "step": 15720
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.5526452861075976e-05,
+      "loss": 1.1898,
+      "step": 15730
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.552347644914056e-05,
+      "loss": 1.1797,
+      "step": 15740
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.552050003720515e-05,
+      "loss": 1.1879,
+      "step": 15750
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.551752362526974e-05,
+      "loss": 1.1908,
+      "step": 15760
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.551454721333433e-05,
+      "loss": 1.1874,
+      "step": 15770
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.551157080139892e-05,
+      "loss": 1.1989,
+      "step": 15780
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.5508594389463504e-05,
+      "loss": 1.1972,
+      "step": 15790
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.550561797752809e-05,
+      "loss": 1.1991,
+      "step": 15800
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.550264156559268e-05,
+      "loss": 1.2048,
+      "step": 15810
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.5499665153657265e-05,
+      "loss": 1.2066,
+      "step": 15820
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.549668874172186e-05,
+      "loss": 1.2009,
+      "step": 15830
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.5493712329786445e-05,
+      "loss": 1.2081,
+      "step": 15840
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.549073591785103e-05,
+      "loss": 1.1987,
+      "step": 15850
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.548775950591562e-05,
+      "loss": 1.2055,
+      "step": 15860
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.548478309398021e-05,
+      "loss": 1.1849,
+      "step": 15870
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.54818066820448e-05,
+      "loss": 1.1996,
+      "step": 15880
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.5478830270109386e-05,
+      "loss": 1.1994,
+      "step": 15890
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.547585385817397e-05,
+      "loss": 1.2023,
+      "step": 15900
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.547287744623856e-05,
+      "loss": 1.201,
+      "step": 15910
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.546990103430315e-05,
+      "loss": 1.2049,
+      "step": 15920
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.546692462236774e-05,
+      "loss": 1.1994,
+      "step": 15930
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.546394821043233e-05,
+      "loss": 1.1945,
+      "step": 15940
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.5460971798496915e-05,
+      "loss": 1.2,
+      "step": 15950
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.54579953865615e-05,
+      "loss": 1.18,
+      "step": 15960
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.5455018974626095e-05,
+      "loss": 1.2056,
+      "step": 15970
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.545204256269068e-05,
+      "loss": 1.1988,
+      "step": 15980
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.544906615075527e-05,
+      "loss": 1.1918,
+      "step": 15990
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.5446089738819856e-05,
+      "loss": 1.2038,
+      "step": 16000
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.544311332688444e-05,
+      "loss": 1.197,
+      "step": 16010
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.544013691494903e-05,
+      "loss": 1.2034,
+      "step": 16020
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.5437160503013617e-05,
+      "loss": 1.1903,
+      "step": 16030
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.543418409107821e-05,
+      "loss": 1.2052,
+      "step": 16040
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.54312076791428e-05,
+      "loss": 1.1866,
+      "step": 16050
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.5428231267207384e-05,
+      "loss": 1.1919,
+      "step": 16060
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.542525485527197e-05,
+      "loss": 1.1984,
+      "step": 16070
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.5422278443336565e-05,
+      "loss": 1.1892,
+      "step": 16080
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.541930203140115e-05,
+      "loss": 1.2101,
+      "step": 16090
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.541632561946574e-05,
+      "loss": 1.1961,
+      "step": 16100
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.5413349207530325e-05,
+      "loss": 1.189,
+      "step": 16110
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.541037279559491e-05,
+      "loss": 1.2025,
+      "step": 16120
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.54073963836595e-05,
+      "loss": 1.1957,
+      "step": 16130
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.5404419971724086e-05,
+      "loss": 1.2099,
+      "step": 16140
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.540144355978868e-05,
+      "loss": 1.1879,
+      "step": 16150
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.5398467147853267e-05,
+      "loss": 1.1947,
+      "step": 16160
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.5395490735917854e-05,
+      "loss": 1.194,
+      "step": 16170
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.539251432398244e-05,
+      "loss": 1.1904,
+      "step": 16180
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.5389537912047034e-05,
+      "loss": 1.1986,
+      "step": 16190
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.538656150011162e-05,
+      "loss": 1.185,
+      "step": 16200
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.53835850881762e-05,
+      "loss": 1.1831,
+      "step": 16210
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.5380608676240795e-05,
+      "loss": 1.1865,
+      "step": 16220
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.537763226430538e-05,
+      "loss": 1.1927,
+      "step": 16230
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.537465585236997e-05,
+      "loss": 1.2009,
+      "step": 16240
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.5371679440434555e-05,
+      "loss": 1.2114,
+      "step": 16250
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.536870302849915e-05,
+      "loss": 1.1997,
+      "step": 16260
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.5365726616563736e-05,
+      "loss": 1.1791,
+      "step": 16270
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.536275020462832e-05,
+      "loss": 1.1893,
+      "step": 16280
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.535977379269291e-05,
+      "loss": 1.1896,
+      "step": 16290
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.5356797380757504e-05,
+      "loss": 1.1912,
+      "step": 16300
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.5353820968822084e-05,
+      "loss": 1.1993,
+      "step": 16310
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.535084455688668e-05,
+      "loss": 1.1918,
+      "step": 16320
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.5347868144951264e-05,
+      "loss": 1.1911,
+      "step": 16330
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.534489173301585e-05,
+      "loss": 1.1954,
+      "step": 16340
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.534191532108044e-05,
+      "loss": 1.1835,
+      "step": 16350
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.533893890914503e-05,
+      "loss": 1.1934,
+      "step": 16360
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.533596249720962e-05,
+      "loss": 1.1888,
+      "step": 16370
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.5332986085274205e-05,
+      "loss": 1.1911,
+      "step": 16380
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.533000967333879e-05,
+      "loss": 1.1721,
+      "step": 16390
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.5327033261403386e-05,
+      "loss": 1.196,
+      "step": 16400
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.532405684946797e-05,
+      "loss": 1.1902,
+      "step": 16410
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.532108043753255e-05,
+      "loss": 1.2061,
+      "step": 16420
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.531810402559715e-05,
+      "loss": 1.2041,
+      "step": 16430
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5315127613661734e-05,
+      "loss": 1.1996,
+      "step": 16440
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.531215120172632e-05,
+      "loss": 1.1817,
+      "step": 16450
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.530917478979091e-05,
+      "loss": 1.1891,
+      "step": 16460
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.53061983778555e-05,
+      "loss": 1.1897,
+      "step": 16470
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.530322196592009e-05,
+      "loss": 1.1874,
+      "step": 16480
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5300245553984675e-05,
+      "loss": 1.188,
+      "step": 16490
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.529726914204926e-05,
+      "loss": 1.1956,
+      "step": 16500
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5294292730113855e-05,
+      "loss": 1.2042,
+      "step": 16510
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5291316318178436e-05,
+      "loss": 1.1866,
+      "step": 16520
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.528833990624302e-05,
+      "loss": 1.1906,
+      "step": 16530
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5285363494307616e-05,
+      "loss": 1.1885,
+      "step": 16540
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.52823870823722e-05,
+      "loss": 1.1993,
+      "step": 16550
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.527941067043679e-05,
+      "loss": 1.1954,
+      "step": 16560
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.527643425850138e-05,
+      "loss": 1.1865,
+      "step": 16570
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.527345784656597e-05,
+      "loss": 1.1832,
+      "step": 16580
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.527048143463056e-05,
+      "loss": 1.1976,
+      "step": 16590
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.5267505022695144e-05,
+      "loss": 1.1771,
+      "step": 16600
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.526452861075973e-05,
+      "loss": 1.1912,
+      "step": 16610
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.526155219882432e-05,
+      "loss": 1.1902,
+      "step": 16620
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.5258575786888905e-05,
+      "loss": 1.1854,
+      "step": 16630
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.525559937495349e-05,
+      "loss": 1.1737,
+      "step": 16640
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.5252622963018086e-05,
+      "loss": 1.1943,
+      "step": 16650
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.524964655108267e-05,
+      "loss": 1.2048,
+      "step": 16660
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.524667013914726e-05,
+      "loss": 1.2057,
+      "step": 16670
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.5243693727211846e-05,
+      "loss": 1.1889,
+      "step": 16680
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.524071731527644e-05,
+      "loss": 1.1988,
+      "step": 16690
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.523774090334103e-05,
+      "loss": 1.189,
+      "step": 16700
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.5234764491405614e-05,
+      "loss": 1.1749,
+      "step": 16710
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.52317880794702e-05,
+      "loss": 1.1987,
+      "step": 16720
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.522881166753479e-05,
+      "loss": 1.1836,
+      "step": 16730
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.5225835255599375e-05,
+      "loss": 1.1958,
+      "step": 16740
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.522285884366397e-05,
+      "loss": 1.1768,
+      "step": 16750
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.5219882431728555e-05,
+      "loss": 1.1856,
+      "step": 16760
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.521690601979314e-05,
+      "loss": 1.1938,
+      "step": 16770
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.521392960785773e-05,
+      "loss": 1.1881,
+      "step": 16780
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.521095319592232e-05,
+      "loss": 1.2021,
+      "step": 16790
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.520797678398691e-05,
+      "loss": 1.1891,
+      "step": 16800
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.5205000372051496e-05,
+      "loss": 1.1827,
+      "step": 16810
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.520202396011608e-05,
+      "loss": 1.1849,
+      "step": 16820
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.519904754818067e-05,
+      "loss": 1.1745,
+      "step": 16830
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.519607113624526e-05,
+      "loss": 1.1923,
+      "step": 16840
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.5193094724309844e-05,
+      "loss": 1.1935,
+      "step": 16850
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.519011831237444e-05,
+      "loss": 1.1806,
+      "step": 16860
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.5187141900439025e-05,
+      "loss": 1.1826,
+      "step": 16870
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.518416548850361e-05,
+      "loss": 1.1836,
+      "step": 16880
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.51811890765682e-05,
+      "loss": 1.2089,
+      "step": 16890
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.517821266463279e-05,
+      "loss": 1.1911,
+      "step": 16900
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.517523625269738e-05,
+      "loss": 1.1937,
+      "step": 16910
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.5172259840761966e-05,
+      "loss": 1.1812,
+      "step": 16920
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.516928342882655e-05,
+      "loss": 1.1869,
+      "step": 16930
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.516630701689114e-05,
+      "loss": 1.1949,
+      "step": 16940
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.5163330604955726e-05,
+      "loss": 1.1926,
+      "step": 16950
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.516035419302031e-05,
+      "loss": 1.1855,
+      "step": 16960
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.515737778108491e-05,
+      "loss": 1.1648,
+      "step": 16970
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.5154401369149494e-05,
+      "loss": 1.1806,
+      "step": 16980
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.515142495721408e-05,
+      "loss": 1.1887,
+      "step": 16990
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.514844854527867e-05,
+      "loss": 1.1931,
+      "step": 17000
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.514547213334326e-05,
+      "loss": 1.1896,
+      "step": 17010
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.514249572140785e-05,
+      "loss": 1.2086,
+      "step": 17020
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.5139519309472435e-05,
+      "loss": 1.1942,
+      "step": 17030
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.513654289753702e-05,
+      "loss": 1.1852,
+      "step": 17040
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.513356648560161e-05,
+      "loss": 1.1882,
+      "step": 17050
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.5130590073666196e-05,
+      "loss": 1.203,
+      "step": 17060
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.512761366173078e-05,
+      "loss": 1.1857,
+      "step": 17070
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.5124637249795376e-05,
+      "loss": 1.2064,
+      "step": 17080
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.5121660837859963e-05,
+      "loss": 1.1951,
+      "step": 17090
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.511868442592455e-05,
+      "loss": 1.1932,
+      "step": 17100
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.511570801398914e-05,
+      "loss": 1.2039,
+      "step": 17110
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.511273160205373e-05,
+      "loss": 1.1903,
+      "step": 17120
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.510975519011831e-05,
+      "loss": 1.1813,
+      "step": 17130
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.5106778778182905e-05,
+      "loss": 1.2101,
+      "step": 17140
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.510380236624749e-05,
+      "loss": 1.1822,
+      "step": 17150
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.510082595431208e-05,
+      "loss": 1.1839,
+      "step": 17160
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.5097849542376665e-05,
+      "loss": 1.1822,
+      "step": 17170
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.509487313044126e-05,
+      "loss": 1.2029,
+      "step": 17180
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.5091896718505846e-05,
+      "loss": 1.2039,
+      "step": 17190
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.508892030657043e-05,
+      "loss": 1.1957,
+      "step": 17200
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.508594389463502e-05,
+      "loss": 1.1914,
+      "step": 17210
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.5082967482699613e-05,
+      "loss": 1.2055,
+      "step": 17220
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.50799910707642e-05,
+      "loss": 1.1905,
+      "step": 17230
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.507701465882878e-05,
+      "loss": 1.1857,
+      "step": 17240
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.5074038246893374e-05,
+      "loss": 1.1867,
+      "step": 17250
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.507106183495796e-05,
+      "loss": 1.1882,
+      "step": 17260
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.506808542302255e-05,
+      "loss": 1.2062,
+      "step": 17270
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.5065109011087135e-05,
+      "loss": 1.1838,
+      "step": 17280
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.506213259915173e-05,
+      "loss": 1.1848,
+      "step": 17290
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.5059156187216315e-05,
+      "loss": 1.1831,
+      "step": 17300
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.50561797752809e-05,
+      "loss": 1.1861,
+      "step": 17310
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.505320336334549e-05,
+      "loss": 1.1731,
+      "step": 17320
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.505022695141008e-05,
+      "loss": 1.1814,
+      "step": 17330
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.504725053947466e-05,
+      "loss": 1.2008,
+      "step": 17340
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.504427412753925e-05,
+      "loss": 1.2027,
+      "step": 17350
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.5041297715603844e-05,
+      "loss": 1.1849,
+      "step": 17360
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.503832130366843e-05,
+      "loss": 1.1857,
+      "step": 17370
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.503534489173302e-05,
+      "loss": 1.1834,
+      "step": 17380
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.5032368479797604e-05,
+      "loss": 1.1909,
+      "step": 17390
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.50293920678622e-05,
+      "loss": 1.1975,
+      "step": 17400
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.5026415655926785e-05,
+      "loss": 1.1899,
+      "step": 17410
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.502343924399137e-05,
+      "loss": 1.2091,
+      "step": 17420
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.502046283205596e-05,
+      "loss": 1.1957,
+      "step": 17430
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.5017486420120546e-05,
+      "loss": 1.1933,
+      "step": 17440
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.501451000818513e-05,
+      "loss": 1.1939,
+      "step": 17450
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.501153359624972e-05,
+      "loss": 1.1777,
+      "step": 17460
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.500855718431431e-05,
+      "loss": 1.1782,
+      "step": 17470
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.50055807723789e-05,
+      "loss": 1.1857,
+      "step": 17480
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.500260436044349e-05,
+      "loss": 1.1837,
+      "step": 17490
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.4999627948508074e-05,
+      "loss": 1.1838,
+      "step": 17500
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.499665153657267e-05,
+      "loss": 1.1935,
+      "step": 17510
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.4993675124637254e-05,
+      "loss": 1.2069,
+      "step": 17520
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.499069871270184e-05,
+      "loss": 1.185,
+      "step": 17530
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.498772230076643e-05,
+      "loss": 1.1975,
+      "step": 17540
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.4984745888831015e-05,
+      "loss": 1.1635,
+      "step": 17550
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.49817694768956e-05,
+      "loss": 1.1897,
+      "step": 17560
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.4978793064960196e-05,
+      "loss": 1.194,
+      "step": 17570
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.497581665302478e-05,
+      "loss": 1.1841,
+      "step": 17580
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.497284024108937e-05,
+      "loss": 1.1692,
+      "step": 17590
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.4969863829153956e-05,
+      "loss": 1.1642,
+      "step": 17600
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.496688741721855e-05,
+      "loss": 1.18,
+      "step": 17610
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.496391100528314e-05,
+      "loss": 1.1878,
+      "step": 17620
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.4960934593347724e-05,
+      "loss": 1.1778,
+      "step": 17630
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.495795818141231e-05,
+      "loss": 1.173,
+      "step": 17640
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.49549817694769e-05,
+      "loss": 1.1848,
+      "step": 17650
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.4952005357541484e-05,
+      "loss": 1.1738,
+      "step": 17660
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.494902894560607e-05,
+      "loss": 1.1857,
+      "step": 17670
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.4946052533670665e-05,
+      "loss": 1.1846,
+      "step": 17680
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.494307612173525e-05,
+      "loss": 1.1998,
+      "step": 17690
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.494009970979984e-05,
+      "loss": 1.1795,
+      "step": 17700
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.4937123297864426e-05,
+      "loss": 1.1759,
+      "step": 17710
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.493414688592902e-05,
+      "loss": 1.1922,
+      "step": 17720
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.4931170473993606e-05,
+      "loss": 1.1809,
+      "step": 17730
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.492819406205819e-05,
+      "loss": 1.1781,
+      "step": 17740
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.492521765012278e-05,
+      "loss": 1.1809,
+      "step": 17750
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.492224123818737e-05,
+      "loss": 1.1704,
+      "step": 17760
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.4919264826251954e-05,
+      "loss": 1.1768,
+      "step": 17770
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.491628841431654e-05,
+      "loss": 1.1972,
+      "step": 17780
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.4913312002381134e-05,
+      "loss": 1.1821,
+      "step": 17790
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.491033559044572e-05,
+      "loss": 1.1947,
+      "step": 17800
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.490735917851031e-05,
+      "loss": 1.1816,
+      "step": 17810
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.4904382766574895e-05,
+      "loss": 1.1794,
+      "step": 17820
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.490140635463949e-05,
+      "loss": 1.188,
+      "step": 17830
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.4898429942704076e-05,
+      "loss": 1.1855,
+      "step": 17840
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.489545353076866e-05,
+      "loss": 1.1694,
+      "step": 17850
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.489247711883325e-05,
+      "loss": 1.188,
+      "step": 17860
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.4889500706897836e-05,
+      "loss": 1.175,
+      "step": 17870
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.488652429496242e-05,
+      "loss": 1.1805,
+      "step": 17880
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.488354788302701e-05,
+      "loss": 1.1933,
+      "step": 17890
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.4880571471091604e-05,
+      "loss": 1.1737,
+      "step": 17900
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.487759505915619e-05,
+      "loss": 1.1799,
+      "step": 17910
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.487461864722078e-05,
+      "loss": 1.1846,
+      "step": 17920
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.4871642235285365e-05,
+      "loss": 1.1953,
+      "step": 17930
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.486866582334996e-05,
+      "loss": 1.1752,
+      "step": 17940
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.486568941141454e-05,
+      "loss": 1.1877,
+      "step": 17950
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.486271299947913e-05,
+      "loss": 1.1805,
+      "step": 17960
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.485973658754372e-05,
+      "loss": 1.1886,
+      "step": 17970
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.4856760175608306e-05,
+      "loss": 1.1831,
+      "step": 17980
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.485378376367289e-05,
+      "loss": 1.1868,
+      "step": 17990
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.4850807351737486e-05,
+      "loss": 1.1897,
+      "step": 18000
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.484783093980207e-05,
+      "loss": 1.1917,
+      "step": 18010
+    },
+    {
+      "epoch": 4.0,
+      "eval_cer": 4.771367521367521,
+      "eval_loss": 1.1306734085083008,
+      "eval_runtime": 5.7447,
+      "eval_samples_per_second": 1.741,
+      "eval_steps_per_second": 0.174,
+      "eval_wer": 1.0389610389610389,
+      "step": 18012
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.484485452786666e-05,
+      "loss": 1.1876,
+      "step": 18020
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.484187811593125e-05,
+      "loss": 1.1865,
+      "step": 18030
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.483890170399584e-05,
+      "loss": 1.1602,
+      "step": 18040
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.483592529206042e-05,
+      "loss": 1.188,
+      "step": 18050
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.483294888012501e-05,
+      "loss": 1.1749,
+      "step": 18060
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.48299724681896e-05,
+      "loss": 1.1685,
+      "step": 18070
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.482699605625419e-05,
+      "loss": 1.1732,
+      "step": 18080
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.4824019644318775e-05,
+      "loss": 1.1896,
+      "step": 18090
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.482104323238336e-05,
+      "loss": 1.201,
+      "step": 18100
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.4818066820447956e-05,
+      "loss": 1.1737,
+      "step": 18110
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.481509040851254e-05,
+      "loss": 1.1684,
+      "step": 18120
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.481211399657713e-05,
+      "loss": 1.181,
+      "step": 18130
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.4809137584641717e-05,
+      "loss": 1.1805,
+      "step": 18140
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.480616117270631e-05,
+      "loss": 1.1936,
+      "step": 18150
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.480318476077089e-05,
+      "loss": 1.1701,
+      "step": 18160
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.480020834883548e-05,
+      "loss": 1.1658,
+      "step": 18170
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.479723193690007e-05,
+      "loss": 1.1796,
+      "step": 18180
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.479425552496466e-05,
+      "loss": 1.1673,
+      "step": 18190
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.4791279113029245e-05,
+      "loss": 1.1847,
+      "step": 18200
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.478830270109383e-05,
+      "loss": 1.1691,
+      "step": 18210
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.4785326289158425e-05,
+      "loss": 1.1889,
+      "step": 18220
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.478234987722301e-05,
+      "loss": 1.1862,
+      "step": 18230
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.47793734652876e-05,
+      "loss": 1.1714,
+      "step": 18240
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.4776397053352186e-05,
+      "loss": 1.1812,
+      "step": 18250
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.477342064141677e-05,
+      "loss": 1.1725,
+      "step": 18260
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.477044422948136e-05,
+      "loss": 1.1828,
+      "step": 18270
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.4767467817545953e-05,
+      "loss": 1.1789,
+      "step": 18280
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.476449140561054e-05,
+      "loss": 1.1832,
+      "step": 18290
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.476151499367513e-05,
+      "loss": 1.1984,
+      "step": 18300
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.4758538581739714e-05,
+      "loss": 1.191,
+      "step": 18310
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.47555621698043e-05,
+      "loss": 1.196,
+      "step": 18320
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.4752585757868895e-05,
+      "loss": 1.1766,
+      "step": 18330
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.474960934593348e-05,
+      "loss": 1.1789,
+      "step": 18340
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.474663293399807e-05,
+      "loss": 1.1829,
+      "step": 18350
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.4743656522062655e-05,
+      "loss": 1.1991,
+      "step": 18360
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.474068011012724e-05,
+      "loss": 1.1808,
+      "step": 18370
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.473770369819183e-05,
+      "loss": 1.1953,
+      "step": 18380
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.473472728625642e-05,
+      "loss": 1.1929,
+      "step": 18390
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.473175087432101e-05,
+      "loss": 1.1723,
+      "step": 18400
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.47287744623856e-05,
+      "loss": 1.1731,
+      "step": 18410
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.4725798050450184e-05,
+      "loss": 1.1768,
+      "step": 18420
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.472282163851478e-05,
+      "loss": 1.186,
+      "step": 18430
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.4719845226579364e-05,
+      "loss": 1.1789,
+      "step": 18440
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.471686881464395e-05,
+      "loss": 1.1774,
+      "step": 18450
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.471389240270854e-05,
+      "loss": 1.167,
+      "step": 18460
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.4710915990773125e-05,
+      "loss": 1.1831,
+      "step": 18470
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.470793957883771e-05,
+      "loss": 1.1712,
+      "step": 18480
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.47049631669023e-05,
+      "loss": 1.1895,
+      "step": 18490
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.470198675496689e-05,
+      "loss": 1.1795,
+      "step": 18500
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.469901034303148e-05,
+      "loss": 1.1727,
+      "step": 18510
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.4696033931096066e-05,
+      "loss": 1.1629,
+      "step": 18520
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.469305751916065e-05,
+      "loss": 1.1801,
+      "step": 18530
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.469008110722525e-05,
+      "loss": 1.1808,
+      "step": 18540
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.4687104695289834e-05,
+      "loss": 1.1783,
+      "step": 18550
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.468412828335442e-05,
+      "loss": 1.1733,
+      "step": 18560
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.468115187141901e-05,
+      "loss": 1.184,
+      "step": 18570
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.4678175459483594e-05,
+      "loss": 1.1799,
+      "step": 18580
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.467519904754818e-05,
+      "loss": 1.1948,
+      "step": 18590
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.467222263561277e-05,
+      "loss": 1.1727,
+      "step": 18600
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.466924622367736e-05,
+      "loss": 1.1636,
+      "step": 18610
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.466626981174195e-05,
+      "loss": 1.1751,
+      "step": 18620
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.4663293399806536e-05,
+      "loss": 1.1761,
+      "step": 18630
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.466031698787112e-05,
+      "loss": 1.1644,
+      "step": 18640
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.4657340575935716e-05,
+      "loss": 1.1815,
+      "step": 18650
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.46543641640003e-05,
+      "loss": 1.1808,
+      "step": 18660
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.465138775206489e-05,
+      "loss": 1.181,
+      "step": 18670
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.464841134012948e-05,
+      "loss": 1.1871,
+      "step": 18680
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.4645434928194064e-05,
+      "loss": 1.1819,
+      "step": 18690
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.464245851625865e-05,
+      "loss": 1.1756,
+      "step": 18700
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.4639482104323244e-05,
+      "loss": 1.1766,
+      "step": 18710
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.463650569238783e-05,
+      "loss": 1.1692,
+      "step": 18720
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.463352928045242e-05,
+      "loss": 1.1795,
+      "step": 18730
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.4630552868517005e-05,
+      "loss": 1.1775,
+      "step": 18740
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.462757645658159e-05,
+      "loss": 1.1786,
+      "step": 18750
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.4624600044646186e-05,
+      "loss": 1.1699,
+      "step": 18760
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.4621623632710766e-05,
+      "loss": 1.1837,
+      "step": 18770
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.461864722077536e-05,
+      "loss": 1.1697,
+      "step": 18780
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.4615670808839946e-05,
+      "loss": 1.1645,
+      "step": 18790
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.461269439690453e-05,
+      "loss": 1.1632,
+      "step": 18800
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.460971798496912e-05,
+      "loss": 1.1807,
+      "step": 18810
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.4606741573033714e-05,
+      "loss": 1.1851,
+      "step": 18820
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.46037651610983e-05,
+      "loss": 1.1804,
+      "step": 18830
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.460078874916289e-05,
+      "loss": 1.1865,
+      "step": 18840
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.4597812337227474e-05,
+      "loss": 1.1751,
+      "step": 18850
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.459483592529207e-05,
+      "loss": 1.1726,
+      "step": 18860
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.459185951335665e-05,
+      "loss": 1.1711,
+      "step": 18870
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.4588883101421235e-05,
+      "loss": 1.1968,
+      "step": 18880
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.458590668948583e-05,
+      "loss": 1.1674,
+      "step": 18890
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.4582930277550416e-05,
+      "loss": 1.1705,
+      "step": 18900
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.4579953865615e-05,
+      "loss": 1.1946,
+      "step": 18910
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.457697745367959e-05,
+      "loss": 1.1768,
+      "step": 18920
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.457400104174418e-05,
+      "loss": 1.1726,
+      "step": 18930
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.457102462980877e-05,
+      "loss": 1.1732,
+      "step": 18940
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.456804821787336e-05,
+      "loss": 1.1964,
+      "step": 18950
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.4565071805937944e-05,
+      "loss": 1.1706,
+      "step": 18960
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.456209539400254e-05,
+      "loss": 1.1793,
+      "step": 18970
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.455911898206712e-05,
+      "loss": 1.1698,
+      "step": 18980
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.4556142570131705e-05,
+      "loss": 1.1704,
+      "step": 18990
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.45531661581963e-05,
+      "loss": 1.1647,
+      "step": 19000
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.4550189746260885e-05,
+      "loss": 1.1889,
+      "step": 19010
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.454721333432547e-05,
+      "loss": 1.1707,
+      "step": 19020
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.454423692239006e-05,
+      "loss": 1.17,
+      "step": 19030
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.454126051045465e-05,
+      "loss": 1.1674,
+      "step": 19040
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.453828409851924e-05,
+      "loss": 1.1845,
+      "step": 19050
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.4535307686583826e-05,
+      "loss": 1.1788,
+      "step": 19060
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.453233127464841e-05,
+      "loss": 1.1638,
+      "step": 19070
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.4529354862713e-05,
+      "loss": 1.1909,
+      "step": 19080
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.452637845077759e-05,
+      "loss": 1.1632,
+      "step": 19090
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.452340203884218e-05,
+      "loss": 1.1759,
+      "step": 19100
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.452042562690677e-05,
+      "loss": 1.1771,
+      "step": 19110
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.4517449214971355e-05,
+      "loss": 1.1818,
+      "step": 19120
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.451447280303594e-05,
+      "loss": 1.1624,
+      "step": 19130
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.4511496391100535e-05,
+      "loss": 1.1744,
+      "step": 19140
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.450851997916512e-05,
+      "loss": 1.1707,
+      "step": 19150
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.450554356722971e-05,
+      "loss": 1.1754,
+      "step": 19160
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 3.4502567155294296e-05,
+      "loss": 1.1679,
+      "step": 19170
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 3.449959074335888e-05,
+      "loss": 1.1775,
+      "step": 19180
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 3.449661433142347e-05,
+      "loss": 1.1674,
+      "step": 19190
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 3.4493637919488057e-05,
+      "loss": 1.1838,
+      "step": 19200
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 3.449066150755265e-05,
+      "loss": 1.169,
+      "step": 19210
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 3.448768509561724e-05,
+      "loss": 1.1851,
+      "step": 19220
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 3.4484708683681824e-05,
+      "loss": 1.1688,
+      "step": 19230
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 3.448173227174641e-05,
+      "loss": 1.1615,
+      "step": 19240
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 3.4478755859811005e-05,
+      "loss": 1.1908,
+      "step": 19250
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.447577944787559e-05,
+      "loss": 1.1756,
+      "step": 19260
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.447280303594018e-05,
+      "loss": 1.1748,
+      "step": 19270
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.4469826624004765e-05,
+      "loss": 1.1755,
+      "step": 19280
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.446685021206935e-05,
+      "loss": 1.1608,
+      "step": 19290
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.446387380013394e-05,
+      "loss": 1.1552,
+      "step": 19300
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.4460897388198526e-05,
+      "loss": 1.1696,
+      "step": 19310
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.445792097626312e-05,
+      "loss": 1.1824,
+      "step": 19320
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.4454944564327707e-05,
+      "loss": 1.1642,
+      "step": 19330
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.4451968152392293e-05,
+      "loss": 1.1704,
+      "step": 19340
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.444899174045688e-05,
+      "loss": 1.1752,
+      "step": 19350
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.4446015328521474e-05,
+      "loss": 1.1658,
+      "step": 19360
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.444303891658606e-05,
+      "loss": 1.1767,
+      "step": 19370
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.444006250465064e-05,
+      "loss": 1.1934,
+      "step": 19380
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.4437086092715235e-05,
+      "loss": 1.1613,
+      "step": 19390
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.443410968077982e-05,
+      "loss": 1.1673,
+      "step": 19400
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.443113326884441e-05,
+      "loss": 1.1737,
+      "step": 19410
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.4428156856908995e-05,
+      "loss": 1.1716,
+      "step": 19420
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.442518044497359e-05,
+      "loss": 1.1662,
+      "step": 19430
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.4422204033038176e-05,
+      "loss": 1.1708,
+      "step": 19440
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.441922762110276e-05,
+      "loss": 1.1701,
+      "step": 19450
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.441625120916735e-05,
+      "loss": 1.1694,
+      "step": 19460
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.4413274797231943e-05,
+      "loss": 1.1693,
+      "step": 19470
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.441029838529653e-05,
+      "loss": 1.1683,
+      "step": 19480
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.440732197336112e-05,
+      "loss": 1.1715,
+      "step": 19490
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.4404345561425704e-05,
+      "loss": 1.1773,
+      "step": 19500
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.440136914949029e-05,
+      "loss": 1.1767,
+      "step": 19510
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.439839273755488e-05,
+      "loss": 1.1708,
+      "step": 19520
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.439541632561947e-05,
+      "loss": 1.1694,
+      "step": 19530
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.439243991368406e-05,
+      "loss": 1.1663,
+      "step": 19540
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.4389463501748645e-05,
+      "loss": 1.1723,
+      "step": 19550
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.438648708981323e-05,
+      "loss": 1.1858,
+      "step": 19560
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.4383510677877826e-05,
+      "loss": 1.1757,
+      "step": 19570
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.438053426594241e-05,
+      "loss": 1.1613,
+      "step": 19580
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.437755785400699e-05,
+      "loss": 1.1707,
+      "step": 19590
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.437458144207159e-05,
+      "loss": 1.1518,
+      "step": 19600
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.4371605030136174e-05,
+      "loss": 1.1791,
+      "step": 19610
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.436862861820076e-05,
+      "loss": 1.1797,
+      "step": 19620
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.436565220626535e-05,
+      "loss": 1.1757,
+      "step": 19630
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.436267579432994e-05,
+      "loss": 1.1853,
+      "step": 19640
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.435969938239453e-05,
+      "loss": 1.1791,
+      "step": 19650
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.4356722970459115e-05,
+      "loss": 1.1723,
+      "step": 19660
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.43537465585237e-05,
+      "loss": 1.1691,
+      "step": 19670
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.4350770146588295e-05,
+      "loss": 1.1587,
+      "step": 19680
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.4347793734652876e-05,
+      "loss": 1.1735,
+      "step": 19690
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.434481732271746e-05,
+      "loss": 1.1756,
+      "step": 19700
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.4341840910782056e-05,
+      "loss": 1.1721,
+      "step": 19710
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.433886449884664e-05,
+      "loss": 1.1581,
+      "step": 19720
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.433588808691123e-05,
+      "loss": 1.167,
+      "step": 19730
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.433291167497582e-05,
+      "loss": 1.1686,
+      "step": 19740
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.432993526304041e-05,
+      "loss": 1.1586,
+      "step": 19750
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.4326958851105e-05,
+      "loss": 1.1686,
+      "step": 19760
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.4323982439169584e-05,
+      "loss": 1.1681,
+      "step": 19770
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.432100602723417e-05,
+      "loss": 1.1702,
+      "step": 19780
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.431802961529876e-05,
+      "loss": 1.156,
+      "step": 19790
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.4315053203363345e-05,
+      "loss": 1.1697,
+      "step": 19800
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.431207679142793e-05,
+      "loss": 1.1781,
+      "step": 19810
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.4309100379492526e-05,
+      "loss": 1.1698,
+      "step": 19820
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.430612396755711e-05,
+      "loss": 1.1773,
+      "step": 19830
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.43031475556217e-05,
+      "loss": 1.17,
+      "step": 19840
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.4300171143686286e-05,
+      "loss": 1.1767,
+      "step": 19850
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.429719473175088e-05,
+      "loss": 1.1653,
+      "step": 19860
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.429421831981547e-05,
+      "loss": 1.1655,
+      "step": 19870
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.4291241907880054e-05,
+      "loss": 1.1661,
+      "step": 19880
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.428826549594464e-05,
+      "loss": 1.1701,
+      "step": 19890
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.428528908400923e-05,
+      "loss": 1.1626,
+      "step": 19900
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.4282312672073814e-05,
+      "loss": 1.1675,
+      "step": 19910
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.427933626013841e-05,
+      "loss": 1.1736,
+      "step": 19920
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.4276359848202995e-05,
+      "loss": 1.1588,
+      "step": 19930
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.427338343626758e-05,
+      "loss": 1.1791,
+      "step": 19940
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.427040702433217e-05,
+      "loss": 1.1579,
+      "step": 19950
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.426743061239676e-05,
+      "loss": 1.1736,
+      "step": 19960
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.426445420046135e-05,
+      "loss": 1.1538,
+      "step": 19970
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.4261477788525936e-05,
+      "loss": 1.1696,
+      "step": 19980
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.425850137659052e-05,
+      "loss": 1.1579,
+      "step": 19990
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.425552496465511e-05,
+      "loss": 1.1591,
+      "step": 20000
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.42525485527197e-05,
+      "loss": 1.1802,
+      "step": 20010
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.4249572140784284e-05,
+      "loss": 1.1778,
+      "step": 20020
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.424659572884888e-05,
+      "loss": 1.1851,
+      "step": 20030
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.4243619316913464e-05,
+      "loss": 1.1631,
+      "step": 20040
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.424064290497805e-05,
+      "loss": 1.1571,
+      "step": 20050
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.423766649304264e-05,
+      "loss": 1.1596,
+      "step": 20060
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.423469008110723e-05,
+      "loss": 1.1552,
+      "step": 20070
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.423171366917182e-05,
+      "loss": 1.1715,
+      "step": 20080
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.4228737257236406e-05,
+      "loss": 1.1604,
+      "step": 20090
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.422576084530099e-05,
+      "loss": 1.1717,
+      "step": 20100
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.422278443336558e-05,
+      "loss": 1.1684,
+      "step": 20110
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.4219808021430166e-05,
+      "loss": 1.1566,
+      "step": 20120
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.421683160949475e-05,
+      "loss": 1.1836,
+      "step": 20130
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.421385519755935e-05,
+      "loss": 1.1687,
+      "step": 20140
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.4210878785623934e-05,
+      "loss": 1.1715,
+      "step": 20150
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.420790237368852e-05,
+      "loss": 1.1726,
+      "step": 20160
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.420492596175311e-05,
+      "loss": 1.1541,
+      "step": 20170
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.42019495498177e-05,
+      "loss": 1.1746,
+      "step": 20180
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.419897313788229e-05,
+      "loss": 1.1821,
+      "step": 20190
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.419599672594687e-05,
+      "loss": 1.1599,
+      "step": 20200
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.419302031401146e-05,
+      "loss": 1.1647,
+      "step": 20210
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.419004390207605e-05,
+      "loss": 1.1622,
+      "step": 20220
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.4187067490140636e-05,
+      "loss": 1.175,
+      "step": 20230
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.418409107820522e-05,
+      "loss": 1.1709,
+      "step": 20240
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.4181114666269816e-05,
+      "loss": 1.1749,
+      "step": 20250
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.41781382543344e-05,
+      "loss": 1.1683,
+      "step": 20260
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.417516184239899e-05,
+      "loss": 1.1703,
+      "step": 20270
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.417218543046358e-05,
+      "loss": 1.1715,
+      "step": 20280
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.416920901852817e-05,
+      "loss": 1.1696,
+      "step": 20290
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.416623260659276e-05,
+      "loss": 1.1645,
+      "step": 20300
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.4163256194657345e-05,
+      "loss": 1.1626,
+      "step": 20310
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.416027978272193e-05,
+      "loss": 1.1765,
+      "step": 20320
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.415730337078652e-05,
+      "loss": 1.1737,
+      "step": 20330
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.4154326958851105e-05,
+      "loss": 1.1885,
+      "step": 20340
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.41513505469157e-05,
+      "loss": 1.1606,
+      "step": 20350
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.4148374134980286e-05,
+      "loss": 1.1574,
+      "step": 20360
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.414539772304487e-05,
+      "loss": 1.1705,
+      "step": 20370
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.414242131110946e-05,
+      "loss": 1.1627,
+      "step": 20380
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.413944489917405e-05,
+      "loss": 1.1736,
+      "step": 20390
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.413646848723864e-05,
+      "loss": 1.1619,
+      "step": 20400
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.413349207530322e-05,
+      "loss": 1.1779,
+      "step": 20410
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.4130515663367814e-05,
+      "loss": 1.1516,
+      "step": 20420
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.41275392514324e-05,
+      "loss": 1.1565,
+      "step": 20430
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.412456283949699e-05,
+      "loss": 1.1675,
+      "step": 20440
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.4121586427561575e-05,
+      "loss": 1.1688,
+      "step": 20450
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.411861001562617e-05,
+      "loss": 1.1661,
+      "step": 20460
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.4115633603690755e-05,
+      "loss": 1.1707,
+      "step": 20470
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.411265719175534e-05,
+      "loss": 1.1739,
+      "step": 20480
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.410968077981993e-05,
+      "loss": 1.154,
+      "step": 20490
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.410670436788452e-05,
+      "loss": 1.1518,
+      "step": 20500
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.41037279559491e-05,
+      "loss": 1.1657,
+      "step": 20510
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.410075154401369e-05,
+      "loss": 1.1515,
+      "step": 20520
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.4097775132078284e-05,
+      "loss": 1.1588,
+      "step": 20530
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.409479872014287e-05,
+      "loss": 1.1693,
+      "step": 20540
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.409182230820746e-05,
+      "loss": 1.1618,
+      "step": 20550
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.4088845896272044e-05,
+      "loss": 1.1539,
+      "step": 20560
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.408586948433664e-05,
+      "loss": 1.1478,
+      "step": 20570
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.4082893072401225e-05,
+      "loss": 1.165,
+      "step": 20580
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.407991666046581e-05,
+      "loss": 1.1615,
+      "step": 20590
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.40769402485304e-05,
+      "loss": 1.1758,
+      "step": 20600
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.4073963836594985e-05,
+      "loss": 1.145,
+      "step": 20610
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.407098742465957e-05,
+      "loss": 1.1712,
+      "step": 20620
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.406801101272416e-05,
+      "loss": 1.156,
+      "step": 20630
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.406503460078875e-05,
+      "loss": 1.1626,
+      "step": 20640
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.406205818885334e-05,
+      "loss": 1.1486,
+      "step": 20650
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.405908177691793e-05,
+      "loss": 1.1625,
+      "step": 20660
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4056105364982514e-05,
+      "loss": 1.1765,
+      "step": 20670
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.405312895304711e-05,
+      "loss": 1.1678,
+      "step": 20680
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4050152541111694e-05,
+      "loss": 1.1692,
+      "step": 20690
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.404717612917628e-05,
+      "loss": 1.1639,
+      "step": 20700
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.404419971724087e-05,
+      "loss": 1.1538,
+      "step": 20710
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.4041223305305455e-05,
+      "loss": 1.1631,
+      "step": 20720
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.403824689337004e-05,
+      "loss": 1.1629,
+      "step": 20730
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.4035270481434635e-05,
+      "loss": 1.1688,
+      "step": 20740
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.403229406949922e-05,
+      "loss": 1.1798,
+      "step": 20750
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.402931765756381e-05,
+      "loss": 1.1651,
+      "step": 20760
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.4026341245628396e-05,
+      "loss": 1.1632,
+      "step": 20770
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.402336483369299e-05,
+      "loss": 1.1786,
+      "step": 20780
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.402038842175758e-05,
+      "loss": 1.1687,
+      "step": 20790
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.4017412009822164e-05,
+      "loss": 1.1762,
+      "step": 20800
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.401443559788675e-05,
+      "loss": 1.1486,
+      "step": 20810
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.401145918595134e-05,
+      "loss": 1.1645,
+      "step": 20820
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.4008482774015924e-05,
+      "loss": 1.17,
+      "step": 20830
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.400550636208051e-05,
+      "loss": 1.1564,
+      "step": 20840
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.4002529950145105e-05,
+      "loss": 1.1594,
+      "step": 20850
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.399955353820969e-05,
+      "loss": 1.1767,
+      "step": 20860
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.399657712627428e-05,
+      "loss": 1.1704,
+      "step": 20870
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.3993600714338866e-05,
+      "loss": 1.1507,
+      "step": 20880
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.399062430240346e-05,
+      "loss": 1.1579,
+      "step": 20890
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.3987647890468046e-05,
+      "loss": 1.1656,
+      "step": 20900
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.398467147853263e-05,
+      "loss": 1.1598,
+      "step": 20910
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.398169506659722e-05,
+      "loss": 1.1595,
+      "step": 20920
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.397871865466181e-05,
+      "loss": 1.1641,
+      "step": 20930
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.3975742242726394e-05,
+      "loss": 1.161,
+      "step": 20940
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.397276583079098e-05,
+      "loss": 1.1653,
+      "step": 20950
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.3969789418855574e-05,
+      "loss": 1.1663,
+      "step": 20960
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.396681300692016e-05,
+      "loss": 1.163,
+      "step": 20970
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.396383659498475e-05,
+      "loss": 1.1634,
+      "step": 20980
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.3960860183049335e-05,
+      "loss": 1.1775,
+      "step": 20990
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.395788377111393e-05,
+      "loss": 1.1675,
+      "step": 21000
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.3954907359178516e-05,
+      "loss": 1.1597,
+      "step": 21010
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.3951930947243096e-05,
+      "loss": 1.167,
+      "step": 21020
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.394895453530769e-05,
+      "loss": 1.1652,
+      "step": 21030
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.3945978123372276e-05,
+      "loss": 1.1672,
+      "step": 21040
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.394300171143686e-05,
+      "loss": 1.1776,
+      "step": 21050
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.394002529950145e-05,
+      "loss": 1.1585,
+      "step": 21060
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.3937048887566044e-05,
+      "loss": 1.1763,
+      "step": 21070
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.393407247563063e-05,
+      "loss": 1.1698,
+      "step": 21080
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.393109606369522e-05,
+      "loss": 1.175,
+      "step": 21090
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.3928119651759805e-05,
+      "loss": 1.1681,
+      "step": 21100
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.39251432398244e-05,
+      "loss": 1.1733,
+      "step": 21110
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.392216682788898e-05,
+      "loss": 1.1633,
+      "step": 21120
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.391919041595357e-05,
+      "loss": 1.1499,
+      "step": 21130
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.391621400401816e-05,
+      "loss": 1.1702,
+      "step": 21140
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.3913237592082746e-05,
+      "loss": 1.1688,
+      "step": 21150
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.391026118014733e-05,
+      "loss": 1.1714,
+      "step": 21160
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.3907284768211926e-05,
+      "loss": 1.1471,
+      "step": 21170
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.390430835627651e-05,
+      "loss": 1.1679,
+      "step": 21180
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.39013319443411e-05,
+      "loss": 1.1579,
+      "step": 21190
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.389835553240569e-05,
+      "loss": 1.1676,
+      "step": 21200
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.389537912047028e-05,
+      "loss": 1.156,
+      "step": 21210
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.389240270853487e-05,
+      "loss": 1.1729,
+      "step": 21220
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.388942629659945e-05,
+      "loss": 1.1583,
+      "step": 21230
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.388644988466404e-05,
+      "loss": 1.1617,
+      "step": 21240
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.388347347272863e-05,
+      "loss": 1.1568,
+      "step": 21250
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.3880497060793215e-05,
+      "loss": 1.1668,
+      "step": 21260
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.38775206488578e-05,
+      "loss": 1.1518,
+      "step": 21270
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.3874544236922396e-05,
+      "loss": 1.1707,
+      "step": 21280
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.387156782498698e-05,
+      "loss": 1.1532,
+      "step": 21290
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.386859141305157e-05,
+      "loss": 1.154,
+      "step": 21300
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.3865615001116156e-05,
+      "loss": 1.1614,
+      "step": 21310
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.386263858918075e-05,
+      "loss": 1.1709,
+      "step": 21320
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.385966217724533e-05,
+      "loss": 1.1752,
+      "step": 21330
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.385668576530992e-05,
+      "loss": 1.1704,
+      "step": 21340
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.385370935337451e-05,
+      "loss": 1.1442,
+      "step": 21350
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.38507329414391e-05,
+      "loss": 1.1447,
+      "step": 21360
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.3847756529503685e-05,
+      "loss": 1.165,
+      "step": 21370
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.384478011756827e-05,
+      "loss": 1.1536,
+      "step": 21380
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.3841803705632865e-05,
+      "loss": 1.1533,
+      "step": 21390
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.383882729369745e-05,
+      "loss": 1.1403,
+      "step": 21400
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.383585088176204e-05,
+      "loss": 1.1657,
+      "step": 21410
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.3832874469826626e-05,
+      "loss": 1.1614,
+      "step": 21420
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.382989805789121e-05,
+      "loss": 1.1558,
+      "step": 21430
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.38269216459558e-05,
+      "loss": 1.1504,
+      "step": 21440
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.382394523402039e-05,
+      "loss": 1.165,
+      "step": 21450
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.382096882208498e-05,
+      "loss": 1.1629,
+      "step": 21460
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.381799241014957e-05,
+      "loss": 1.1593,
+      "step": 21470
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.3815015998214154e-05,
+      "loss": 1.1652,
+      "step": 21480
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.381203958627874e-05,
+      "loss": 1.1665,
+      "step": 21490
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.3809063174343335e-05,
+      "loss": 1.1583,
+      "step": 21500
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 3.380608676240792e-05,
+      "loss": 1.1689,
+      "step": 21510
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 3.380311035047251e-05,
+      "loss": 1.1625,
+      "step": 21520
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 3.3800133938537095e-05,
+      "loss": 1.1563,
+      "step": 21530
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 3.379715752660168e-05,
+      "loss": 1.1666,
+      "step": 21540
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 3.379418111466627e-05,
+      "loss": 1.1712,
+      "step": 21550
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 3.379120470273086e-05,
+      "loss": 1.1518,
+      "step": 21560
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 3.378822829079545e-05,
+      "loss": 1.1574,
+      "step": 21570
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 3.378525187886004e-05,
+      "loss": 1.1527,
+      "step": 21580
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 3.3782275466924624e-05,
+      "loss": 1.1538,
+      "step": 21590
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 3.377929905498922e-05,
+      "loss": 1.166,
+      "step": 21600
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 3.3776322643053804e-05,
+      "loss": 1.1724,
+      "step": 21610
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 3.377334623111839e-05,
+      "loss": 1.16,
+      "step": 21620
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 3.377036981918298e-05,
+      "loss": 1.1779,
+      "step": 21630
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.3767393407247565e-05,
+      "loss": 1.1547,
+      "step": 21640
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.376441699531215e-05,
+      "loss": 1.1609,
+      "step": 21650
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.376144058337674e-05,
+      "loss": 1.1595,
+      "step": 21660
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.375846417144133e-05,
+      "loss": 1.1668,
+      "step": 21670
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.375548775950592e-05,
+      "loss": 1.1654,
+      "step": 21680
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.3752511347570506e-05,
+      "loss": 1.1546,
+      "step": 21690
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.374953493563509e-05,
+      "loss": 1.1769,
+      "step": 21700
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.374655852369969e-05,
+      "loss": 1.1642,
+      "step": 21710
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.3743582111764274e-05,
+      "loss": 1.181,
+      "step": 21720
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.374060569982886e-05,
+      "loss": 1.1638,
+      "step": 21730
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.373762928789345e-05,
+      "loss": 1.1493,
+      "step": 21740
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.3734652875958034e-05,
+      "loss": 1.1642,
+      "step": 21750
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.373167646402262e-05,
+      "loss": 1.1626,
+      "step": 21760
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.372870005208721e-05,
+      "loss": 1.1516,
+      "step": 21770
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.37257236401518e-05,
+      "loss": 1.1547,
+      "step": 21780
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.372274722821639e-05,
+      "loss": 1.1502,
+      "step": 21790
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.3719770816280976e-05,
+      "loss": 1.1574,
+      "step": 21800
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.371679440434556e-05,
+      "loss": 1.1575,
+      "step": 21810
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.3713817992410156e-05,
+      "loss": 1.1446,
+      "step": 21820
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.371084158047474e-05,
+      "loss": 1.1827,
+      "step": 21830
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.370786516853933e-05,
+      "loss": 1.1777,
+      "step": 21840
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.370488875660392e-05,
+      "loss": 1.1527,
+      "step": 21850
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.3701912344668504e-05,
+      "loss": 1.15,
+      "step": 21860
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.369893593273309e-05,
+      "loss": 1.1619,
+      "step": 21870
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.369595952079768e-05,
+      "loss": 1.1691,
+      "step": 21880
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.369298310886227e-05,
+      "loss": 1.1766,
+      "step": 21890
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.369000669692686e-05,
+      "loss": 1.1617,
+      "step": 21900
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.3687030284991445e-05,
+      "loss": 1.1659,
+      "step": 21910
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.368405387305603e-05,
+      "loss": 1.157,
+      "step": 21920
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.3681077461120626e-05,
+      "loss": 1.1624,
+      "step": 21930
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.3678101049185206e-05,
+      "loss": 1.1662,
+      "step": 21940
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.36751246372498e-05,
+      "loss": 1.1439,
+      "step": 21950
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.3672148225314386e-05,
+      "loss": 1.1533,
+      "step": 21960
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.366917181337897e-05,
+      "loss": 1.1605,
+      "step": 21970
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.366619540144356e-05,
+      "loss": 1.1663,
+      "step": 21980
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.3663218989508154e-05,
+      "loss": 1.1563,
+      "step": 21990
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.366024257757274e-05,
+      "loss": 1.1674,
+      "step": 22000
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.365726616563733e-05,
+      "loss": 1.1707,
+      "step": 22010
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.3654289753701914e-05,
+      "loss": 1.168,
+      "step": 22020
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.365131334176651e-05,
+      "loss": 1.1596,
+      "step": 22030
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.3648336929831095e-05,
+      "loss": 1.158,
+      "step": 22040
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.3645360517895675e-05,
+      "loss": 1.1365,
+      "step": 22050
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.364238410596027e-05,
+      "loss": 1.155,
+      "step": 22060
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.3639407694024856e-05,
+      "loss": 1.1576,
+      "step": 22070
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.363643128208944e-05,
+      "loss": 1.1541,
+      "step": 22080
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.363345487015403e-05,
+      "loss": 1.1624,
+      "step": 22090
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.363047845821862e-05,
+      "loss": 1.1842,
+      "step": 22100
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.362750204628321e-05,
+      "loss": 1.1597,
+      "step": 22110
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.36245256343478e-05,
+      "loss": 1.1577,
+      "step": 22120
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.3621549222412384e-05,
+      "loss": 1.158,
+      "step": 22130
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.361857281047698e-05,
+      "loss": 1.1535,
+      "step": 22140
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.361559639854156e-05,
+      "loss": 1.1613,
+      "step": 22150
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.3612619986606145e-05,
+      "loss": 1.1599,
+      "step": 22160
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.360964357467074e-05,
+      "loss": 1.1409,
+      "step": 22170
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.3606667162735325e-05,
+      "loss": 1.166,
+      "step": 22180
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.360369075079991e-05,
+      "loss": 1.1574,
+      "step": 22190
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.36007143388645e-05,
+      "loss": 1.1538,
+      "step": 22200
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.359773792692909e-05,
+      "loss": 1.1596,
+      "step": 22210
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.359476151499368e-05,
+      "loss": 1.1556,
+      "step": 22220
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.3591785103058266e-05,
+      "loss": 1.142,
+      "step": 22230
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.358880869112285e-05,
+      "loss": 1.1517,
+      "step": 22240
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.358583227918744e-05,
+      "loss": 1.1535,
+      "step": 22250
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.358285586725203e-05,
+      "loss": 1.1592,
+      "step": 22260
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.357987945531662e-05,
+      "loss": 1.1488,
+      "step": 22270
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.357690304338121e-05,
+      "loss": 1.16,
+      "step": 22280
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.3573926631445795e-05,
+      "loss": 1.158,
+      "step": 22290
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.357095021951038e-05,
+      "loss": 1.1436,
+      "step": 22300
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.356797380757497e-05,
+      "loss": 1.1768,
+      "step": 22310
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.356499739563956e-05,
+      "loss": 1.1643,
+      "step": 22320
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.356202098370415e-05,
+      "loss": 1.1461,
+      "step": 22330
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.3559044571768736e-05,
+      "loss": 1.166,
+      "step": 22340
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.355606815983332e-05,
+      "loss": 1.1469,
+      "step": 22350
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.355309174789791e-05,
+      "loss": 1.159,
+      "step": 22360
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.3550115335962497e-05,
+      "loss": 1.1619,
+      "step": 22370
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.354713892402709e-05,
+      "loss": 1.1669,
+      "step": 22380
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.354416251209168e-05,
+      "loss": 1.1526,
+      "step": 22390
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.3541186100156264e-05,
+      "loss": 1.1698,
+      "step": 22400
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.353820968822085e-05,
+      "loss": 1.153,
+      "step": 22410
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.3535233276285445e-05,
+      "loss": 1.1518,
+      "step": 22420
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.353225686435003e-05,
+      "loss": 1.1611,
+      "step": 22430
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.352928045241462e-05,
+      "loss": 1.1491,
+      "step": 22440
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.3526304040479205e-05,
+      "loss": 1.1524,
+      "step": 22450
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.352332762854379e-05,
+      "loss": 1.166,
+      "step": 22460
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.352035121660838e-05,
+      "loss": 1.1317,
+      "step": 22470
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.3517374804672966e-05,
+      "loss": 1.1352,
+      "step": 22480
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.351439839273756e-05,
+      "loss": 1.1552,
+      "step": 22490
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.3511421980802147e-05,
+      "loss": 1.1499,
+      "step": 22500
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.3508445568866733e-05,
+      "loss": 1.162,
+      "step": 22510
+    },
+    {
+      "epoch": 5.0,
+      "eval_cer": 4.760683760683761,
+      "eval_loss": 1.123809576034546,
+      "eval_runtime": 5.6237,
+      "eval_samples_per_second": 1.778,
+      "eval_steps_per_second": 0.178,
+      "eval_wer": 0.987012987012987,
+      "step": 22515
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.350546915693132e-05,
+      "loss": 1.1466,
+      "step": 22520
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.3502492744995914e-05,
+      "loss": 1.1569,
+      "step": 22530
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.34995163330605e-05,
+      "loss": 1.158,
+      "step": 22540
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.349653992112509e-05,
+      "loss": 1.1414,
+      "step": 22550
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.3493563509189675e-05,
+      "loss": 1.1489,
+      "step": 22560
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.349058709725426e-05,
+      "loss": 1.1484,
+      "step": 22570
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.348761068531885e-05,
+      "loss": 1.1474,
+      "step": 22580
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.3484634273383435e-05,
+      "loss": 1.1479,
+      "step": 22590
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.348165786144803e-05,
+      "loss": 1.1599,
+      "step": 22600
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.3478681449512616e-05,
+      "loss": 1.158,
+      "step": 22610
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.34757050375772e-05,
+      "loss": 1.1658,
+      "step": 22620
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.347272862564179e-05,
+      "loss": 1.1425,
+      "step": 22630
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.3469752213706383e-05,
+      "loss": 1.1537,
+      "step": 22640
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.346677580177097e-05,
+      "loss": 1.1634,
+      "step": 22650
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.346379938983556e-05,
+      "loss": 1.1563,
+      "step": 22660
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.3460822977900144e-05,
+      "loss": 1.1562,
+      "step": 22670
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.345784656596473e-05,
+      "loss": 1.1524,
+      "step": 22680
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.345487015402932e-05,
+      "loss": 1.155,
+      "step": 22690
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.345189374209391e-05,
+      "loss": 1.1502,
+      "step": 22700
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.34489173301585e-05,
+      "loss": 1.1406,
+      "step": 22710
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.3445940918223085e-05,
+      "loss": 1.1537,
+      "step": 22720
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.344296450628767e-05,
+      "loss": 1.1635,
+      "step": 22730
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.343998809435226e-05,
+      "loss": 1.1565,
+      "step": 22740
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.343701168241685e-05,
+      "loss": 1.1613,
+      "step": 22750
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.343403527048143e-05,
+      "loss": 1.1417,
+      "step": 22760
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.343105885854603e-05,
+      "loss": 1.1454,
+      "step": 22770
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.3428082446610614e-05,
+      "loss": 1.1471,
+      "step": 22780
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.34251060346752e-05,
+      "loss": 1.1529,
+      "step": 22790
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.342212962273979e-05,
+      "loss": 1.1473,
+      "step": 22800
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.341915321080438e-05,
+      "loss": 1.1526,
+      "step": 22810
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.341617679886897e-05,
+      "loss": 1.1469,
+      "step": 22820
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.3413200386933555e-05,
+      "loss": 1.1453,
+      "step": 22830
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.341022397499814e-05,
+      "loss": 1.1496,
+      "step": 22840
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.3407247563062735e-05,
+      "loss": 1.1428,
+      "step": 22850
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 3.3404271151127316e-05,
+      "loss": 1.1599,
+      "step": 22860
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 3.34012947391919e-05,
+      "loss": 1.146,
+      "step": 22870
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 3.3398318327256496e-05,
+      "loss": 1.162,
+      "step": 22880
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 3.339534191532108e-05,
+      "loss": 1.1636,
+      "step": 22890
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.339236550338567e-05,
+      "loss": 1.1418,
+      "step": 22900
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.338938909145026e-05,
+      "loss": 1.1663,
+      "step": 22910
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.338641267951485e-05,
+      "loss": 1.1332,
+      "step": 22920
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.338343626757944e-05,
+      "loss": 1.1467,
+      "step": 22930
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.3380459855644024e-05,
+      "loss": 1.1534,
+      "step": 22940
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.337748344370861e-05,
+      "loss": 1.1384,
+      "step": 22950
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.3374507031773205e-05,
+      "loss": 1.1498,
+      "step": 22960
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.3371530619837785e-05,
+      "loss": 1.1464,
+      "step": 22970
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.336855420790237e-05,
+      "loss": 1.1468,
+      "step": 22980
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.3365577795966966e-05,
+      "loss": 1.1502,
+      "step": 22990
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.336260138403155e-05,
+      "loss": 1.1702,
+      "step": 23000
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.335962497209614e-05,
+      "loss": 1.1491,
+      "step": 23010
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.3356648560160726e-05,
+      "loss": 1.1614,
+      "step": 23020
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.335367214822532e-05,
+      "loss": 1.1513,
+      "step": 23030
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 3.335069573628991e-05,
+      "loss": 1.1492,
+      "step": 23040
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 3.3347719324354494e-05,
+      "loss": 1.1478,
+      "step": 23050
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 3.334474291241908e-05,
+      "loss": 1.1497,
+      "step": 23060
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 3.334176650048367e-05,
+      "loss": 1.1429,
+      "step": 23070
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.3338790088548254e-05,
+      "loss": 1.1526,
+      "step": 23080
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.333581367661285e-05,
+      "loss": 1.1459,
+      "step": 23090
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.3332837264677435e-05,
+      "loss": 1.1522,
+      "step": 23100
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.332986085274202e-05,
+      "loss": 1.1507,
+      "step": 23110
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.332688444080661e-05,
+      "loss": 1.1683,
+      "step": 23120
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.33239080288712e-05,
+      "loss": 1.1394,
+      "step": 23130
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.332093161693579e-05,
+      "loss": 1.1531,
+      "step": 23140
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.3317955205000376e-05,
+      "loss": 1.1453,
+      "step": 23150
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.331497879306496e-05,
+      "loss": 1.1517,
+      "step": 23160
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.331200238112955e-05,
+      "loss": 1.1467,
+      "step": 23170
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.330902596919414e-05,
+      "loss": 1.1421,
+      "step": 23180
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.3306049557258724e-05,
+      "loss": 1.1574,
+      "step": 23190
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.330307314532332e-05,
+      "loss": 1.1487,
+      "step": 23200
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.3300096733387904e-05,
+      "loss": 1.1495,
+      "step": 23210
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.329712032145249e-05,
+      "loss": 1.145,
+      "step": 23220
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.329414390951708e-05,
+      "loss": 1.1373,
+      "step": 23230
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.329116749758167e-05,
+      "loss": 1.1467,
+      "step": 23240
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.328819108564626e-05,
+      "loss": 1.1649,
+      "step": 23250
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.3285214673710846e-05,
+      "loss": 1.1664,
+      "step": 23260
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.328223826177543e-05,
+      "loss": 1.1659,
+      "step": 23270
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.327926184984002e-05,
+      "loss": 1.1564,
+      "step": 23280
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.3276285437904606e-05,
+      "loss": 1.1481,
+      "step": 23290
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.327330902596919e-05,
+      "loss": 1.1443,
+      "step": 23300
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.327033261403379e-05,
+      "loss": 1.1386,
+      "step": 23310
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.3267356202098374e-05,
+      "loss": 1.1568,
+      "step": 23320
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.326437979016296e-05,
+      "loss": 1.165,
+      "step": 23330
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.326140337822755e-05,
+      "loss": 1.1387,
+      "step": 23340
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3.325842696629214e-05,
+      "loss": 1.1544,
+      "step": 23350
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3.325545055435673e-05,
+      "loss": 1.1365,
+      "step": 23360
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3.3252474142421315e-05,
+      "loss": 1.1515,
+      "step": 23370
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3.32494977304859e-05,
+      "loss": 1.1353,
+      "step": 23380
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3.324652131855049e-05,
+      "loss": 1.1478,
+      "step": 23390
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.3243544906615076e-05,
+      "loss": 1.1512,
+      "step": 23400
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.324056849467966e-05,
+      "loss": 1.1315,
+      "step": 23410
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.3237592082744256e-05,
+      "loss": 1.1498,
+      "step": 23420
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.323461567080884e-05,
+      "loss": 1.1529,
+      "step": 23430
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.323163925887343e-05,
+      "loss": 1.1572,
+      "step": 23440
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.322866284693802e-05,
+      "loss": 1.1334,
+      "step": 23450
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.322568643500261e-05,
+      "loss": 1.1499,
+      "step": 23460
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.32227100230672e-05,
+      "loss": 1.1518,
+      "step": 23470
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.3219733611131785e-05,
+      "loss": 1.1435,
+      "step": 23480
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.321675719919637e-05,
+      "loss": 1.1403,
+      "step": 23490
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.321378078726096e-05,
+      "loss": 1.1495,
+      "step": 23500
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.3210804375325545e-05,
+      "loss": 1.1517,
+      "step": 23510
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.320782796339014e-05,
+      "loss": 1.1477,
+      "step": 23520
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.3204851551454726e-05,
+      "loss": 1.1585,
+      "step": 23530
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.320187513951931e-05,
+      "loss": 1.1474,
+      "step": 23540
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.31988987275839e-05,
+      "loss": 1.1525,
+      "step": 23550
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.319592231564849e-05,
+      "loss": 1.1645,
+      "step": 23560
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.319294590371308e-05,
+      "loss": 1.1541,
+      "step": 23570
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.318996949177766e-05,
+      "loss": 1.1635,
+      "step": 23580
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.3186993079842254e-05,
+      "loss": 1.1409,
+      "step": 23590
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.318401666790684e-05,
+      "loss": 1.1324,
+      "step": 23600
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.318104025597143e-05,
+      "loss": 1.1481,
+      "step": 23610
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.3178063844036015e-05,
+      "loss": 1.1433,
+      "step": 23620
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.317508743210061e-05,
+      "loss": 1.1332,
+      "step": 23630
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.3172111020165195e-05,
+      "loss": 1.1467,
+      "step": 23640
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.316913460822978e-05,
+      "loss": 1.1349,
+      "step": 23650
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.316615819629437e-05,
+      "loss": 1.1424,
+      "step": 23660
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.316318178435896e-05,
+      "loss": 1.1449,
+      "step": 23670
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.316020537242354e-05,
+      "loss": 1.1366,
+      "step": 23680
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.315722896048813e-05,
+      "loss": 1.1534,
+      "step": 23690
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.3154252548552723e-05,
+      "loss": 1.1406,
+      "step": 23700
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.315127613661731e-05,
+      "loss": 1.1421,
+      "step": 23710
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.31482997246819e-05,
+      "loss": 1.1393,
+      "step": 23720
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.3145323312746484e-05,
+      "loss": 1.1589,
+      "step": 23730
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.314234690081108e-05,
+      "loss": 1.1349,
+      "step": 23740
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.3139370488875665e-05,
+      "loss": 1.144,
+      "step": 23750
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.313639407694025e-05,
+      "loss": 1.1503,
+      "step": 23760
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.313341766500484e-05,
+      "loss": 1.1512,
+      "step": 23770
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.313044125306943e-05,
+      "loss": 1.1453,
+      "step": 23780
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.312746484113401e-05,
+      "loss": 1.1563,
+      "step": 23790
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.31244884291986e-05,
+      "loss": 1.1443,
+      "step": 23800
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.312151201726319e-05,
+      "loss": 1.1469,
+      "step": 23810
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.311853560532778e-05,
+      "loss": 1.1503,
+      "step": 23820
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.311555919339237e-05,
+      "loss": 1.1593,
+      "step": 23830
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.3112582781456954e-05,
+      "loss": 1.1611,
+      "step": 23840
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.310960636952155e-05,
+      "loss": 1.1406,
+      "step": 23850
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.3106629957586134e-05,
+      "loss": 1.1429,
+      "step": 23860
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.310365354565072e-05,
+      "loss": 1.1507,
+      "step": 23870
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.310067713371531e-05,
+      "loss": 1.1395,
+      "step": 23880
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.3097700721779895e-05,
+      "loss": 1.1448,
+      "step": 23890
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.309472430984448e-05,
+      "loss": 1.1411,
+      "step": 23900
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.3091747897909075e-05,
+      "loss": 1.1485,
+      "step": 23910
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.308877148597366e-05,
+      "loss": 1.1451,
+      "step": 23920
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.308579507403825e-05,
+      "loss": 1.1598,
+      "step": 23930
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.3082818662102836e-05,
+      "loss": 1.1606,
+      "step": 23940
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.307984225016743e-05,
+      "loss": 1.1415,
+      "step": 23950
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.307686583823202e-05,
+      "loss": 1.1454,
+      "step": 23960
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.3073889426296604e-05,
+      "loss": 1.1499,
+      "step": 23970
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.307091301436119e-05,
+      "loss": 1.1466,
+      "step": 23980
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.306793660242578e-05,
+      "loss": 1.1426,
+      "step": 23990
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.3064960190490364e-05,
+      "loss": 1.1378,
+      "step": 24000
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.306198377855495e-05,
+      "loss": 1.1533,
+      "step": 24010
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.3059007366619545e-05,
+      "loss": 1.1346,
+      "step": 24020
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.305603095468413e-05,
+      "loss": 1.1448,
+      "step": 24030
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.305305454274872e-05,
+      "loss": 1.1493,
+      "step": 24040
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.3050078130813306e-05,
+      "loss": 1.1428,
+      "step": 24050
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.30471017188779e-05,
+      "loss": 1.1474,
+      "step": 24060
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.3044125306942486e-05,
+      "loss": 1.1419,
+      "step": 24070
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.304114889500707e-05,
+      "loss": 1.1325,
+      "step": 24080
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.303817248307166e-05,
+      "loss": 1.1349,
+      "step": 24090
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.303519607113625e-05,
+      "loss": 1.1408,
+      "step": 24100
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.3032219659200834e-05,
+      "loss": 1.1342,
+      "step": 24110
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.302924324726542e-05,
+      "loss": 1.1291,
+      "step": 24120
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.3026266835330014e-05,
+      "loss": 1.1342,
+      "step": 24130
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.30232904233946e-05,
+      "loss": 1.1467,
+      "step": 24140
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.302031401145919e-05,
+      "loss": 1.1495,
+      "step": 24150
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.3017337599523775e-05,
+      "loss": 1.1502,
+      "step": 24160
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.301436118758837e-05,
+      "loss": 1.1529,
+      "step": 24170
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.3011384775652956e-05,
+      "loss": 1.1453,
+      "step": 24180
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.300840836371754e-05,
+      "loss": 1.1261,
+      "step": 24190
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.300543195178213e-05,
+      "loss": 1.1442,
+      "step": 24200
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.3002455539846716e-05,
+      "loss": 1.1393,
+      "step": 24210
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.29994791279113e-05,
+      "loss": 1.1455,
+      "step": 24220
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.299650271597589e-05,
+      "loss": 1.135,
+      "step": 24230
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.2993526304040484e-05,
+      "loss": 1.145,
+      "step": 24240
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.299054989210507e-05,
+      "loss": 1.1435,
+      "step": 24250
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.298757348016966e-05,
+      "loss": 1.1418,
+      "step": 24260
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.2984597068234244e-05,
+      "loss": 1.1444,
+      "step": 24270
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.298162065629884e-05,
+      "loss": 1.1418,
+      "step": 24280
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.2978644244363425e-05,
+      "loss": 1.14,
+      "step": 24290
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.297566783242801e-05,
+      "loss": 1.1295,
+      "step": 24300
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.29726914204926e-05,
+      "loss": 1.1561,
+      "step": 24310
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.2969715008557186e-05,
+      "loss": 1.1518,
+      "step": 24320
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.296673859662177e-05,
+      "loss": 1.1416,
+      "step": 24330
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.2963762184686366e-05,
+      "loss": 1.1267,
+      "step": 24340
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.296078577275095e-05,
+      "loss": 1.1377,
+      "step": 24350
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.295780936081554e-05,
+      "loss": 1.1279,
+      "step": 24360
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.295483294888013e-05,
+      "loss": 1.1476,
+      "step": 24370
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.295185653694472e-05,
+      "loss": 1.1429,
+      "step": 24380
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.294888012500931e-05,
+      "loss": 1.1525,
+      "step": 24390
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.294590371307389e-05,
+      "loss": 1.1424,
+      "step": 24400
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.294292730113848e-05,
+      "loss": 1.1388,
+      "step": 24410
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.293995088920307e-05,
+      "loss": 1.137,
+      "step": 24420
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3.2936974477267655e-05,
+      "loss": 1.1454,
+      "step": 24430
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3.293399806533224e-05,
+      "loss": 1.1497,
+      "step": 24440
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3.2931021653396836e-05,
+      "loss": 1.1455,
+      "step": 24450
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3.292804524146142e-05,
+      "loss": 1.1531,
+      "step": 24460
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3.292506882952601e-05,
+      "loss": 1.1431,
+      "step": 24470
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.2922092417590596e-05,
+      "loss": 1.1339,
+      "step": 24480
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.291911600565519e-05,
+      "loss": 1.1541,
+      "step": 24490
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.291613959371977e-05,
+      "loss": 1.1429,
+      "step": 24500
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.291316318178436e-05,
+      "loss": 1.1486,
+      "step": 24510
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3.291018676984895e-05,
+      "loss": 1.1346,
+      "step": 24520
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3.290721035791354e-05,
+      "loss": 1.1436,
+      "step": 24530
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3.2904233945978125e-05,
+      "loss": 1.1375,
+      "step": 24540
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3.290125753404271e-05,
+      "loss": 1.15,
+      "step": 24550
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3.2898281122107305e-05,
+      "loss": 1.1295,
+      "step": 24560
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3.289530471017189e-05,
+      "loss": 1.1475,
+      "step": 24570
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3.289232829823648e-05,
+      "loss": 1.1519,
+      "step": 24580
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3.2889351886301066e-05,
+      "loss": 1.1467,
+      "step": 24590
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3.288637547436565e-05,
+      "loss": 1.1412,
+      "step": 24600
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.288339906243024e-05,
+      "loss": 1.1405,
+      "step": 24610
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.2880422650494827e-05,
+      "loss": 1.1493,
+      "step": 24620
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.287744623855942e-05,
+      "loss": 1.1379,
+      "step": 24630
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.287446982662401e-05,
+      "loss": 1.1475,
+      "step": 24640
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.2871493414688594e-05,
+      "loss": 1.1465,
+      "step": 24650
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.286851700275318e-05,
+      "loss": 1.1303,
+      "step": 24660
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.2865540590817775e-05,
+      "loss": 1.1598,
+      "step": 24670
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.286256417888236e-05,
+      "loss": 1.147,
+      "step": 24680
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.285958776694695e-05,
+      "loss": 1.1545,
+      "step": 24690
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.2856611355011535e-05,
+      "loss": 1.1434,
+      "step": 24700
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.285363494307612e-05,
+      "loss": 1.1378,
+      "step": 24710
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.285065853114071e-05,
+      "loss": 1.1525,
+      "step": 24720
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.28476821192053e-05,
+      "loss": 1.1447,
+      "step": 24730
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.284470570726989e-05,
+      "loss": 1.1455,
+      "step": 24740
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.2841729295334477e-05,
+      "loss": 1.1493,
+      "step": 24750
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.2838752883399064e-05,
+      "loss": 1.1325,
+      "step": 24760
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.283577647146366e-05,
+      "loss": 1.1465,
+      "step": 24770
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.2832800059528244e-05,
+      "loss": 1.1466,
+      "step": 24780
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.282982364759283e-05,
+      "loss": 1.1436,
+      "step": 24790
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.282684723565742e-05,
+      "loss": 1.1375,
+      "step": 24800
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.2823870823722005e-05,
+      "loss": 1.13,
+      "step": 24810
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.282089441178659e-05,
+      "loss": 1.1535,
+      "step": 24820
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.281791799985118e-05,
+      "loss": 1.1394,
+      "step": 24830
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.281494158791577e-05,
+      "loss": 1.139,
+      "step": 24840
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.281196517598036e-05,
+      "loss": 1.1451,
+      "step": 24850
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.2808988764044946e-05,
+      "loss": 1.1322,
+      "step": 24860
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.280601235210953e-05,
+      "loss": 1.148,
+      "step": 24870
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.280303594017413e-05,
+      "loss": 1.1346,
+      "step": 24880
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.2800059528238714e-05,
+      "loss": 1.1374,
+      "step": 24890
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.27970831163033e-05,
+      "loss": 1.1467,
+      "step": 24900
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.279410670436789e-05,
+      "loss": 1.1378,
+      "step": 24910
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.2791130292432474e-05,
+      "loss": 1.1526,
+      "step": 24920
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.278815388049706e-05,
+      "loss": 1.1319,
+      "step": 24930
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.278517746856165e-05,
+      "loss": 1.1481,
+      "step": 24940
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.278220105662624e-05,
+      "loss": 1.1386,
+      "step": 24950
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.277922464469083e-05,
+      "loss": 1.1466,
+      "step": 24960
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.2776248232755415e-05,
+      "loss": 1.1442,
+      "step": 24970
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.277327182082e-05,
+      "loss": 1.1437,
+      "step": 24980
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.2770295408884596e-05,
+      "loss": 1.1311,
+      "step": 24990
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.276731899694918e-05,
+      "loss": 1.152,
+      "step": 25000
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.276434258501376e-05,
+      "loss": 1.1618,
+      "step": 25010
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.276136617307836e-05,
+      "loss": 1.1416,
+      "step": 25020
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.2758389761142944e-05,
+      "loss": 1.1492,
+      "step": 25030
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.275541334920753e-05,
+      "loss": 1.1258,
+      "step": 25040
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.275243693727212e-05,
+      "loss": 1.1429,
+      "step": 25050
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.274946052533671e-05,
+      "loss": 1.148,
+      "step": 25060
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.27464841134013e-05,
+      "loss": 1.1222,
+      "step": 25070
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.2743507701465885e-05,
+      "loss": 1.1405,
+      "step": 25080
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.274053128953047e-05,
+      "loss": 1.1338,
+      "step": 25090
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.2737554877595066e-05,
+      "loss": 1.1301,
+      "step": 25100
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.273457846565965e-05,
+      "loss": 1.142,
+      "step": 25110
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.273160205372424e-05,
+      "loss": 1.1309,
+      "step": 25120
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.2728625641788826e-05,
+      "loss": 1.1475,
+      "step": 25130
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.272564922985341e-05,
+      "loss": 1.1472,
+      "step": 25140
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.2722672817918e-05,
+      "loss": 1.1492,
+      "step": 25150
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.2719696405982594e-05,
+      "loss": 1.149,
+      "step": 25160
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.271671999404718e-05,
+      "loss": 1.1451,
+      "step": 25170
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.271374358211177e-05,
+      "loss": 1.1492,
+      "step": 25180
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.2710767170176354e-05,
+      "loss": 1.1458,
+      "step": 25190
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.270779075824095e-05,
+      "loss": 1.1466,
+      "step": 25200
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.2704814346305535e-05,
+      "loss": 1.1472,
+      "step": 25210
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.2701837934370115e-05,
+      "loss": 1.1394,
+      "step": 25220
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.269886152243471e-05,
+      "loss": 1.1519,
+      "step": 25230
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 3.2695885110499296e-05,
+      "loss": 1.1419,
+      "step": 25240
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 3.269290869856388e-05,
+      "loss": 1.1484,
+      "step": 25250
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 3.268993228662847e-05,
+      "loss": 1.1515,
+      "step": 25260
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 3.268695587469306e-05,
+      "loss": 1.1321,
+      "step": 25270
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 3.268397946275765e-05,
+      "loss": 1.1344,
+      "step": 25280
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 3.268100305082224e-05,
+      "loss": 1.1419,
+      "step": 25290
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 3.2678026638886824e-05,
+      "loss": 1.1533,
+      "step": 25300
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 3.267505022695142e-05,
+      "loss": 1.1442,
+      "step": 25310
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 3.2672073815016e-05,
+      "loss": 1.1375,
+      "step": 25320
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 3.2669097403080585e-05,
+      "loss": 1.1324,
+      "step": 25330
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 3.266612099114518e-05,
+      "loss": 1.1409,
+      "step": 25340
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 3.2663144579209765e-05,
+      "loss": 1.1286,
+      "step": 25350
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 3.266016816727435e-05,
+      "loss": 1.1402,
+      "step": 25360
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 3.265719175533894e-05,
+      "loss": 1.1396,
+      "step": 25370
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 3.265421534340353e-05,
+      "loss": 1.1485,
+      "step": 25380
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 3.265123893146812e-05,
+      "loss": 1.1469,
+      "step": 25390
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 3.2648262519532706e-05,
+      "loss": 1.1315,
+      "step": 25400
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 3.264528610759729e-05,
+      "loss": 1.1512,
+      "step": 25410
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 3.264230969566188e-05,
+      "loss": 1.1417,
+      "step": 25420
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 3.263933328372647e-05,
+      "loss": 1.1385,
+      "step": 25430
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 3.2636356871791054e-05,
+      "loss": 1.1272,
+      "step": 25440
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 3.263338045985565e-05,
+      "loss": 1.1459,
+      "step": 25450
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 3.2630404047920235e-05,
+      "loss": 1.1447,
+      "step": 25460
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 3.262742763598482e-05,
+      "loss": 1.1436,
+      "step": 25470
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 3.262445122404941e-05,
+      "loss": 1.1187,
+      "step": 25480
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 3.2621474812114e-05,
+      "loss": 1.1383,
+      "step": 25490
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 3.261849840017859e-05,
+      "loss": 1.141,
+      "step": 25500
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 3.2615521988243176e-05,
+      "loss": 1.1364,
+      "step": 25510
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 3.261254557630776e-05,
+      "loss": 1.1502,
+      "step": 25520
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 3.260956916437235e-05,
+      "loss": 1.1421,
+      "step": 25530
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 3.2606592752436936e-05,
+      "loss": 1.1454,
+      "step": 25540
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 3.260361634050153e-05,
+      "loss": 1.1566,
+      "step": 25550
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 3.260063992856612e-05,
+      "loss": 1.1455,
+      "step": 25560
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 3.2597663516630704e-05,
+      "loss": 1.1538,
+      "step": 25570
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 3.259468710469529e-05,
+      "loss": 1.1433,
+      "step": 25580
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 3.2591710692759885e-05,
+      "loss": 1.1392,
+      "step": 25590
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 3.258873428082447e-05,
+      "loss": 1.134,
+      "step": 25600
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 3.258575786888906e-05,
+      "loss": 1.1431,
+      "step": 25610
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 3.2582781456953645e-05,
+      "loss": 1.1279,
+      "step": 25620
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 3.257980504501823e-05,
+      "loss": 1.1406,
+      "step": 25630
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 3.257682863308282e-05,
+      "loss": 1.152,
+      "step": 25640
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 3.2573852221147406e-05,
+      "loss": 1.1403,
+      "step": 25650
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 3.2570875809212e-05,
+      "loss": 1.1275,
+      "step": 25660
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 3.2567899397276586e-05,
+      "loss": 1.1439,
+      "step": 25670
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 3.2564922985341173e-05,
+      "loss": 1.1347,
+      "step": 25680
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 3.256194657340576e-05,
+      "loss": 1.1463,
+      "step": 25690
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 3.2558970161470354e-05,
+      "loss": 1.1596,
+      "step": 25700
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 3.255599374953494e-05,
+      "loss": 1.1264,
+      "step": 25710
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 3.255301733759953e-05,
+      "loss": 1.1388,
+      "step": 25720
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 3.2550040925664115e-05,
+      "loss": 1.1313,
+      "step": 25730
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 3.25470645137287e-05,
+      "loss": 1.1268,
+      "step": 25740
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 3.254408810179329e-05,
+      "loss": 1.1379,
+      "step": 25750
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 3.2541111689857875e-05,
+      "loss": 1.1401,
+      "step": 25760
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 3.253813527792247e-05,
+      "loss": 1.1391,
+      "step": 25770
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 3.2535158865987056e-05,
+      "loss": 1.134,
+      "step": 25780
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 3.253218245405164e-05,
+      "loss": 1.1424,
+      "step": 25790
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 3.252920604211623e-05,
+      "loss": 1.1363,
+      "step": 25800
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 3.2526229630180823e-05,
+      "loss": 1.1428,
+      "step": 25810
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 3.252325321824541e-05,
+      "loss": 1.1312,
+      "step": 25820
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 3.252027680631e-05,
+      "loss": 1.1397,
+      "step": 25830
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 3.2517300394374584e-05,
+      "loss": 1.125,
+      "step": 25840
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 3.251432398243917e-05,
+      "loss": 1.1336,
+      "step": 25850
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 3.251134757050376e-05,
+      "loss": 1.1512,
+      "step": 25860
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 3.2508371158568345e-05,
+      "loss": 1.1278,
+      "step": 25870
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 3.250539474663294e-05,
+      "loss": 1.1375,
+      "step": 25880
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 3.2502418334697525e-05,
+      "loss": 1.1283,
+      "step": 25890
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 3.249944192276211e-05,
+      "loss": 1.1205,
+      "step": 25900
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 3.24964655108267e-05,
+      "loss": 1.1407,
+      "step": 25910
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 3.249348909889129e-05,
+      "loss": 1.1447,
+      "step": 25920
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 3.249051268695588e-05,
+      "loss": 1.1346,
+      "step": 25930
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 3.248753627502047e-05,
+      "loss": 1.1344,
+      "step": 25940
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 3.2484559863085054e-05,
+      "loss": 1.1448,
+      "step": 25950
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 3.248158345114964e-05,
+      "loss": 1.141,
+      "step": 25960
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 3.247860703921423e-05,
+      "loss": 1.1292,
+      "step": 25970
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 3.247563062727882e-05,
+      "loss": 1.1615,
+      "step": 25980
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 3.247265421534341e-05,
+      "loss": 1.1543,
+      "step": 25990
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 3.2469677803407995e-05,
+      "loss": 1.1468,
+      "step": 26000
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 3.246670139147258e-05,
+      "loss": 1.1542,
+      "step": 26010
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 3.2463724979537175e-05,
+      "loss": 1.145,
+      "step": 26020
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 3.246074856760176e-05,
+      "loss": 1.1433,
+      "step": 26030
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 3.245777215566634e-05,
+      "loss": 1.1616,
+      "step": 26040
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 3.2454795743730936e-05,
+      "loss": 1.1212,
+      "step": 26050
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 3.245181933179552e-05,
+      "loss": 1.141,
+      "step": 26060
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 3.244884291986011e-05,
+      "loss": 1.1478,
+      "step": 26070
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 3.24458665079247e-05,
+      "loss": 1.1525,
+      "step": 26080
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 3.244289009598929e-05,
+      "loss": 1.1381,
+      "step": 26090
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 3.243991368405388e-05,
+      "loss": 1.1238,
+      "step": 26100
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 3.2436937272118464e-05,
+      "loss": 1.1579,
+      "step": 26110
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 3.243396086018305e-05,
+      "loss": 1.1415,
+      "step": 26120
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 3.2430984448247645e-05,
+      "loss": 1.1498,
+      "step": 26130
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 3.2428008036312225e-05,
+      "loss": 1.1348,
+      "step": 26140
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 3.242503162437681e-05,
+      "loss": 1.1541,
+      "step": 26150
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 3.2422055212441406e-05,
+      "loss": 1.1497,
+      "step": 26160
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 3.241907880050599e-05,
+      "loss": 1.1275,
+      "step": 26170
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 3.241610238857058e-05,
+      "loss": 1.1324,
+      "step": 26180
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 3.2413125976635166e-05,
+      "loss": 1.1374,
+      "step": 26190
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 3.241014956469976e-05,
+      "loss": 1.1427,
+      "step": 26200
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 3.240717315276435e-05,
+      "loss": 1.1306,
+      "step": 26210
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 3.2404196740828934e-05,
+      "loss": 1.1316,
+      "step": 26220
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.240122032889352e-05,
+      "loss": 1.1281,
+      "step": 26230
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.239824391695811e-05,
+      "loss": 1.1327,
+      "step": 26240
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.2395267505022694e-05,
+      "loss": 1.1512,
+      "step": 26250
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.239229109308729e-05,
+      "loss": 1.1423,
+      "step": 26260
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.2389314681151875e-05,
+      "loss": 1.1442,
+      "step": 26270
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.238633826921646e-05,
+      "loss": 1.1433,
+      "step": 26280
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.238336185728105e-05,
+      "loss": 1.1488,
+      "step": 26290
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.2380385445345636e-05,
+      "loss": 1.1273,
+      "step": 26300
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.237740903341023e-05,
+      "loss": 1.1462,
+      "step": 26310
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.2374432621474816e-05,
+      "loss": 1.1524,
+      "step": 26320
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.23714562095394e-05,
+      "loss": 1.1447,
+      "step": 26330
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.236847979760399e-05,
+      "loss": 1.1559,
+      "step": 26340
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.236550338566858e-05,
+      "loss": 1.1414,
+      "step": 26350
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.2362526973733164e-05,
+      "loss": 1.1408,
+      "step": 26360
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 3.235955056179776e-05,
+      "loss": 1.1624,
+      "step": 26370
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 3.2356574149862344e-05,
+      "loss": 1.1263,
+      "step": 26380
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 3.235359773792693e-05,
+      "loss": 1.1386,
+      "step": 26390
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 3.235062132599152e-05,
+      "loss": 1.1305,
+      "step": 26400
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 3.234764491405611e-05,
+      "loss": 1.1382,
+      "step": 26410
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 3.23446685021207e-05,
+      "loss": 1.1182,
+      "step": 26420
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 3.2341692090185286e-05,
+      "loss": 1.1304,
+      "step": 26430
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 3.233871567824987e-05,
+      "loss": 1.1304,
+      "step": 26440
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 3.233573926631446e-05,
+      "loss": 1.1422,
+      "step": 26450
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 3.2332762854379046e-05,
+      "loss": 1.144,
+      "step": 26460
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 3.232978644244363e-05,
+      "loss": 1.1401,
+      "step": 26470
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 3.232681003050823e-05,
+      "loss": 1.1505,
+      "step": 26480
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 3.2323833618572814e-05,
+      "loss": 1.1402,
+      "step": 26490
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 3.23208572066374e-05,
+      "loss": 1.148,
+      "step": 26500
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 3.231788079470199e-05,
+      "loss": 1.1439,
+      "step": 26510
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 3.231490438276658e-05,
+      "loss": 1.1443,
+      "step": 26520
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 3.231192797083117e-05,
+      "loss": 1.1403,
+      "step": 26530
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 3.2308951558895755e-05,
+      "loss": 1.1186,
+      "step": 26540
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 3.230597514696034e-05,
+      "loss": 1.1331,
+      "step": 26550
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 3.230299873502493e-05,
+      "loss": 1.144,
+      "step": 26560
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 3.2300022323089516e-05,
+      "loss": 1.1367,
+      "step": 26570
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 3.22970459111541e-05,
+      "loss": 1.1398,
+      "step": 26580
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 3.2294069499218696e-05,
+      "loss": 1.1399,
+      "step": 26590
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 3.229109308728328e-05,
+      "loss": 1.1486,
+      "step": 26600
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 3.228811667534787e-05,
+      "loss": 1.1332,
+      "step": 26610
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 3.228514026341246e-05,
+      "loss": 1.1355,
+      "step": 26620
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 3.228216385147705e-05,
+      "loss": 1.1292,
+      "step": 26630
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 3.227918743954164e-05,
+      "loss": 1.1338,
+      "step": 26640
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 3.2276211027606225e-05,
+      "loss": 1.14,
+      "step": 26650
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 3.227323461567081e-05,
+      "loss": 1.1515,
+      "step": 26660
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 3.22702582037354e-05,
+      "loss": 1.133,
+      "step": 26670
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 3.2267281791799985e-05,
+      "loss": 1.1358,
+      "step": 26680
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.226430537986458e-05,
+      "loss": 1.1397,
+      "step": 26690
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.2261328967929166e-05,
+      "loss": 1.1383,
+      "step": 26700
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.225835255599375e-05,
+      "loss": 1.1395,
+      "step": 26710
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.225537614405834e-05,
+      "loss": 1.141,
+      "step": 26720
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.2252399732122927e-05,
+      "loss": 1.1373,
+      "step": 26730
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.224942332018752e-05,
+      "loss": 1.1307,
+      "step": 26740
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.22464469082521e-05,
+      "loss": 1.1406,
+      "step": 26750
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.2243470496316694e-05,
+      "loss": 1.1412,
+      "step": 26760
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.224049408438128e-05,
+      "loss": 1.1403,
+      "step": 26770
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.223751767244587e-05,
+      "loss": 1.1446,
+      "step": 26780
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.2234541260510455e-05,
+      "loss": 1.1498,
+      "step": 26790
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.223156484857505e-05,
+      "loss": 1.1336,
+      "step": 26800
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.2228588436639635e-05,
+      "loss": 1.1526,
+      "step": 26810
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 3.222561202470422e-05,
+      "loss": 1.1347,
+      "step": 26820
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 3.222263561276881e-05,
+      "loss": 1.1406,
+      "step": 26830
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 3.22196592008334e-05,
+      "loss": 1.1319,
+      "step": 26840
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 3.221668278889799e-05,
+      "loss": 1.1241,
+      "step": 26850
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 3.221370637696257e-05,
+      "loss": 1.1377,
+      "step": 26860
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 3.2210729965027163e-05,
+      "loss": 1.1366,
+      "step": 26870
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 3.220775355309175e-05,
+      "loss": 1.127,
+      "step": 26880
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 3.220477714115634e-05,
+      "loss": 1.1366,
+      "step": 26890
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 3.2201800729220924e-05,
+      "loss": 1.1177,
+      "step": 26900
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 3.219882431728552e-05,
+      "loss": 1.1317,
+      "step": 26910
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 3.2195847905350105e-05,
+      "loss": 1.1244,
+      "step": 26920
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 3.219287149341469e-05,
+      "loss": 1.1481,
+      "step": 26930
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 3.218989508147928e-05,
+      "loss": 1.1319,
+      "step": 26940
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 3.218691866954387e-05,
+      "loss": 1.1439,
+      "step": 26950
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3.218394225760845e-05,
+      "loss": 1.1291,
+      "step": 26960
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3.218096584567304e-05,
+      "loss": 1.1307,
+      "step": 26970
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3.217798943373763e-05,
+      "loss": 1.1189,
+      "step": 26980
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3.217501302180222e-05,
+      "loss": 1.1555,
+      "step": 26990
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.217203660986681e-05,
+      "loss": 1.1393,
+      "step": 27000
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.2169060197931394e-05,
+      "loss": 1.1371,
+      "step": 27010
+    },
+    {
+      "epoch": 6.0,
+      "eval_cer": 4.760683760683761,
+      "eval_loss": 1.125946044921875,
+      "eval_runtime": 5.4649,
+      "eval_samples_per_second": 1.83,
+      "eval_steps_per_second": 0.183,
+      "eval_wer": 1.0129870129870129,
+      "step": 27018
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.216608378599599e-05,
+      "loss": 1.1301,
+      "step": 27020
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.2163107374060574e-05,
+      "loss": 1.1482,
+      "step": 27030
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.216013096212516e-05,
+      "loss": 1.1311,
+      "step": 27040
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.215715455018975e-05,
+      "loss": 1.1424,
+      "step": 27050
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.2154178138254335e-05,
+      "loss": 1.1393,
+      "step": 27060
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.215120172631892e-05,
+      "loss": 1.1356,
+      "step": 27070
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.2148225314383515e-05,
+      "loss": 1.1368,
+      "step": 27080
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.21452489024481e-05,
+      "loss": 1.1262,
+      "step": 27090
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.214227249051269e-05,
+      "loss": 1.143,
+      "step": 27100
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.2139296078577276e-05,
+      "loss": 1.1437,
+      "step": 27110
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.213631966664187e-05,
+      "loss": 1.1547,
+      "step": 27120
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 3.213334325470646e-05,
+      "loss": 1.1247,
+      "step": 27130
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.2130366842771044e-05,
+      "loss": 1.1319,
+      "step": 27140
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.212739043083563e-05,
+      "loss": 1.1334,
+      "step": 27150
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.212441401890022e-05,
+      "loss": 1.1383,
+      "step": 27160
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.2121437606964804e-05,
+      "loss": 1.146,
+      "step": 27170
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.211846119502939e-05,
+      "loss": 1.1315,
+      "step": 27180
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.2115484783093985e-05,
+      "loss": 1.1288,
+      "step": 27190
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.211250837115857e-05,
+      "loss": 1.1278,
+      "step": 27200
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.210953195922316e-05,
+      "loss": 1.1257,
+      "step": 27210
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.2106555547287746e-05,
+      "loss": 1.1387,
+      "step": 27220
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.210357913535234e-05,
+      "loss": 1.1211,
+      "step": 27230
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.2100602723416926e-05,
+      "loss": 1.13,
+      "step": 27240
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.209762631148151e-05,
+      "loss": 1.1415,
+      "step": 27250
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 3.20946498995461e-05,
+      "loss": 1.1381,
+      "step": 27260
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.209167348761069e-05,
+      "loss": 1.1333,
+      "step": 27270
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.2088697075675274e-05,
+      "loss": 1.1298,
+      "step": 27280
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.208572066373986e-05,
+      "loss": 1.1334,
+      "step": 27290
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.2082744251804454e-05,
+      "loss": 1.1482,
+      "step": 27300
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.207976783986904e-05,
+      "loss": 1.1305,
+      "step": 27310
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.207679142793363e-05,
+      "loss": 1.124,
+      "step": 27320
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.2073815015998215e-05,
+      "loss": 1.1369,
+      "step": 27330
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.207083860406281e-05,
+      "loss": 1.1349,
+      "step": 27340
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 3.2067862192127396e-05,
+      "loss": 1.1315,
+      "step": 27350
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.206488578019198e-05,
+      "loss": 1.1432,
+      "step": 27360
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.206190936825657e-05,
+      "loss": 1.1341,
+      "step": 27370
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.2058932956321156e-05,
+      "loss": 1.1281,
+      "step": 27380
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.205595654438574e-05,
+      "loss": 1.1307,
+      "step": 27390
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.205298013245033e-05,
+      "loss": 1.1433,
+      "step": 27400
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.2050003720514924e-05,
+      "loss": 1.1423,
+      "step": 27410
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.204702730857951e-05,
+      "loss": 1.1303,
+      "step": 27420
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.20440508966441e-05,
+      "loss": 1.1338,
+      "step": 27430
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 3.2041074484708684e-05,
+      "loss": 1.1475,
+      "step": 27440
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.203809807277328e-05,
+      "loss": 1.1318,
+      "step": 27450
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.2035121660837865e-05,
+      "loss": 1.1147,
+      "step": 27460
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.203214524890245e-05,
+      "loss": 1.1318,
+      "step": 27470
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.202916883696704e-05,
+      "loss": 1.1222,
+      "step": 27480
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.2026192425031626e-05,
+      "loss": 1.1345,
+      "step": 27490
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.202321601309621e-05,
+      "loss": 1.1214,
+      "step": 27500
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.2020239601160806e-05,
+      "loss": 1.1415,
+      "step": 27510
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.201726318922539e-05,
+      "loss": 1.158,
+      "step": 27520
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 3.201428677728998e-05,
+      "loss": 1.1188,
+      "step": 27530
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.201131036535457e-05,
+      "loss": 1.1458,
+      "step": 27540
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.200833395341916e-05,
+      "loss": 1.1312,
+      "step": 27550
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.200535754148375e-05,
+      "loss": 1.1247,
+      "step": 27560
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.200238112954833e-05,
+      "loss": 1.139,
+      "step": 27570
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.199940471761292e-05,
+      "loss": 1.1163,
+      "step": 27580
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.199642830567751e-05,
+      "loss": 1.1281,
+      "step": 27590
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.1993451893742095e-05,
+      "loss": 1.1336,
+      "step": 27600
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.199047548180668e-05,
+      "loss": 1.1264,
+      "step": 27610
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.1987499069871276e-05,
+      "loss": 1.1366,
+      "step": 27620
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.198452265793586e-05,
+      "loss": 1.135,
+      "step": 27630
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.198154624600045e-05,
+      "loss": 1.1409,
+      "step": 27640
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.1978569834065036e-05,
+      "loss": 1.1304,
+      "step": 27650
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.197559342212963e-05,
+      "loss": 1.1209,
+      "step": 27660
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 3.197261701019422e-05,
+      "loss": 1.1341,
+      "step": 27670
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.19696405982588e-05,
+      "loss": 1.1289,
+      "step": 27680
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.196666418632339e-05,
+      "loss": 1.1226,
+      "step": 27690
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.196368777438798e-05,
+      "loss": 1.1317,
+      "step": 27700
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.1960711362452565e-05,
+      "loss": 1.1239,
+      "step": 27710
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.195773495051715e-05,
+      "loss": 1.133,
+      "step": 27720
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.1954758538581745e-05,
+      "loss": 1.1149,
+      "step": 27730
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.195178212664633e-05,
+      "loss": 1.1384,
+      "step": 27740
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.194880571471092e-05,
+      "loss": 1.1254,
+      "step": 27750
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 3.1945829302775506e-05,
+      "loss": 1.1139,
+      "step": 27760
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.19428528908401e-05,
+      "loss": 1.1411,
+      "step": 27770
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.193987647890468e-05,
+      "loss": 1.12,
+      "step": 27780
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.1936900066969267e-05,
+      "loss": 1.1392,
+      "step": 27790
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.193392365503386e-05,
+      "loss": 1.1244,
+      "step": 27800
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.193094724309845e-05,
+      "loss": 1.1241,
+      "step": 27810
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.1927970831163034e-05,
+      "loss": 1.1403,
+      "step": 27820
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.192499441922762e-05,
+      "loss": 1.1135,
+      "step": 27830
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.1922018007292215e-05,
+      "loss": 1.1227,
+      "step": 27840
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 3.19190415953568e-05,
+      "loss": 1.1214,
+      "step": 27850
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.191606518342139e-05,
+      "loss": 1.1382,
+      "step": 27860
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.1913088771485975e-05,
+      "loss": 1.1408,
+      "step": 27870
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.191011235955056e-05,
+      "loss": 1.1362,
+      "step": 27880
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.190713594761515e-05,
+      "loss": 1.1234,
+      "step": 27890
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.190415953567974e-05,
+      "loss": 1.1354,
+      "step": 27900
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.190118312374433e-05,
+      "loss": 1.139,
+      "step": 27910
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.1898206711808917e-05,
+      "loss": 1.13,
+      "step": 27920
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.1895230299873503e-05,
+      "loss": 1.1231,
+      "step": 27930
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 3.18922538879381e-05,
+      "loss": 1.1323,
+      "step": 27940
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.1889277476002684e-05,
+      "loss": 1.1189,
+      "step": 27950
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.188630106406727e-05,
+      "loss": 1.1538,
+      "step": 27960
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.188332465213186e-05,
+      "loss": 1.1212,
+      "step": 27970
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.1880348240196445e-05,
+      "loss": 1.123,
+      "step": 27980
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.187737182826103e-05,
+      "loss": 1.1369,
+      "step": 27990
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.187439541632562e-05,
+      "loss": 1.1362,
+      "step": 28000
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.187141900439021e-05,
+      "loss": 1.1132,
+      "step": 28010
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.18684425924548e-05,
+      "loss": 1.1345,
+      "step": 28020
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.1865466180519386e-05,
+      "loss": 1.1413,
+      "step": 28030
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.186248976858397e-05,
+      "loss": 1.1457,
+      "step": 28040
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.1859513356648567e-05,
+      "loss": 1.1291,
+      "step": 28050
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.1856536944713153e-05,
+      "loss": 1.1315,
+      "step": 28060
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 3.185356053277774e-05,
+      "loss": 1.1286,
+      "step": 28070
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.185058412084233e-05,
+      "loss": 1.1374,
+      "step": 28080
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.1847607708906914e-05,
+      "loss": 1.1317,
+      "step": 28090
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.18446312969715e-05,
+      "loss": 1.119,
+      "step": 28100
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.184165488503609e-05,
+      "loss": 1.1209,
+      "step": 28110
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.183867847310068e-05,
+      "loss": 1.1202,
+      "step": 28120
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.183570206116527e-05,
+      "loss": 1.1234,
+      "step": 28130
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.1832725649229855e-05,
+      "loss": 1.1192,
+      "step": 28140
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.182974923729444e-05,
+      "loss": 1.1176,
+      "step": 28150
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 3.1826772825359036e-05,
+      "loss": 1.112,
+      "step": 28160
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.182379641342362e-05,
+      "loss": 1.101,
+      "step": 28170
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.182082000148821e-05,
+      "loss": 1.1227,
+      "step": 28180
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.18178435895528e-05,
+      "loss": 1.1211,
+      "step": 28190
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.1814867177617384e-05,
+      "loss": 1.1215,
+      "step": 28200
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.181189076568197e-05,
+      "loss": 1.118,
+      "step": 28210
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.180891435374656e-05,
+      "loss": 1.1197,
+      "step": 28220
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.180593794181115e-05,
+      "loss": 1.1226,
+      "step": 28230
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.180296152987574e-05,
+      "loss": 1.1229,
+      "step": 28240
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 3.1799985117940325e-05,
+      "loss": 1.1402,
+      "step": 28250
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.179700870600491e-05,
+      "loss": 1.1353,
+      "step": 28260
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.1794032294069505e-05,
+      "loss": 1.138,
+      "step": 28270
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.179105588213409e-05,
+      "loss": 1.141,
+      "step": 28280
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.178807947019868e-05,
+      "loss": 1.1174,
+      "step": 28290
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.1785103058263266e-05,
+      "loss": 1.1143,
+      "step": 28300
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.178212664632785e-05,
+      "loss": 1.1352,
+      "step": 28310
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.177915023439244e-05,
+      "loss": 1.0983,
+      "step": 28320
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.1776173822457034e-05,
+      "loss": 1.13,
+      "step": 28330
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 3.177319741052162e-05,
+      "loss": 1.1258,
+      "step": 28340
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.177022099858621e-05,
+      "loss": 1.1249,
+      "step": 28350
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.1767244586650794e-05,
+      "loss": 1.1021,
+      "step": 28360
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.176426817471539e-05,
+      "loss": 1.1352,
+      "step": 28370
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.1761291762779975e-05,
+      "loss": 1.1189,
+      "step": 28380
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.1758315350844555e-05,
+      "loss": 1.1306,
+      "step": 28390
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.175533893890915e-05,
+      "loss": 1.135,
+      "step": 28400
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.1752362526973736e-05,
+      "loss": 1.135,
+      "step": 28410
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.174938611503832e-05,
+      "loss": 1.1367,
+      "step": 28420
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 3.174640970310291e-05,
+      "loss": 1.1263,
+      "step": 28430
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.17434332911675e-05,
+      "loss": 1.1311,
+      "step": 28440
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.174045687923209e-05,
+      "loss": 1.1259,
+      "step": 28450
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.173748046729668e-05,
+      "loss": 1.1396,
+      "step": 28460
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.1734504055361264e-05,
+      "loss": 1.1214,
+      "step": 28470
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.173152764342586e-05,
+      "loss": 1.1353,
+      "step": 28480
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.172855123149044e-05,
+      "loss": 1.1185,
+      "step": 28490
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.1725574819555024e-05,
+      "loss": 1.1193,
+      "step": 28500
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.172259840761962e-05,
+      "loss": 1.1177,
+      "step": 28510
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.1719621995684205e-05,
+      "loss": 1.1321,
+      "step": 28520
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.171664558374879e-05,
+      "loss": 1.1201,
+      "step": 28530
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.171366917181338e-05,
+      "loss": 1.1253,
+      "step": 28540
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.171069275987797e-05,
+      "loss": 1.1103,
+      "step": 28550
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.170771634794256e-05,
+      "loss": 1.1142,
+      "step": 28560
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 3.1704739936007146e-05,
+      "loss": 1.1084,
+      "step": 28570
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.170176352407173e-05,
+      "loss": 1.1133,
+      "step": 28580
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.169878711213633e-05,
+      "loss": 1.1138,
+      "step": 28590
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.169581070020091e-05,
+      "loss": 1.1133,
+      "step": 28600
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.1692834288265494e-05,
+      "loss": 1.1275,
+      "step": 28610
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.168985787633009e-05,
+      "loss": 1.1252,
+      "step": 28620
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.1686881464394674e-05,
+      "loss": 1.1296,
+      "step": 28630
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.168390505245926e-05,
+      "loss": 1.1111,
+      "step": 28640
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.168092864052385e-05,
+      "loss": 1.1226,
+      "step": 28650
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 3.167795222858844e-05,
+      "loss": 1.1345,
+      "step": 28660
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.167497581665303e-05,
+      "loss": 1.126,
+      "step": 28670
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.1671999404717616e-05,
+      "loss": 1.1305,
+      "step": 28680
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.16690229927822e-05,
+      "loss": 1.1345,
+      "step": 28690
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.166604658084679e-05,
+      "loss": 1.1291,
+      "step": 28700
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.1663070168911376e-05,
+      "loss": 1.1217,
+      "step": 28710
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.166009375697597e-05,
+      "loss": 1.1175,
+      "step": 28720
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.165711734504056e-05,
+      "loss": 1.1303,
+      "step": 28730
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.1654140933105144e-05,
+      "loss": 1.1199,
+      "step": 28740
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 3.165116452116973e-05,
+      "loss": 1.1226,
+      "step": 28750
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.1648188109234324e-05,
+      "loss": 1.1394,
+      "step": 28760
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.164521169729891e-05,
+      "loss": 1.1399,
+      "step": 28770
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.16422352853635e-05,
+      "loss": 1.1155,
+      "step": 28780
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.1639258873428085e-05,
+      "loss": 1.1284,
+      "step": 28790
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.163628246149267e-05,
+      "loss": 1.1177,
+      "step": 28800
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.163330604955726e-05,
+      "loss": 1.1489,
+      "step": 28810
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.1630329637621846e-05,
+      "loss": 1.1236,
+      "step": 28820
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.162735322568644e-05,
+      "loss": 1.1226,
+      "step": 28830
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 3.1624376813751026e-05,
+      "loss": 1.1155,
+      "step": 28840
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.162140040181561e-05,
+      "loss": 1.1312,
+      "step": 28850
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.16184239898802e-05,
+      "loss": 1.1307,
+      "step": 28860
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.1615447577944794e-05,
+      "loss": 1.1328,
+      "step": 28870
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.161247116600938e-05,
+      "loss": 1.1212,
+      "step": 28880
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.160949475407397e-05,
+      "loss": 1.1313,
+      "step": 28890
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.1606518342138555e-05,
+      "loss": 1.1246,
+      "step": 28900
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.160354193020314e-05,
+      "loss": 1.1133,
+      "step": 28910
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.160056551826773e-05,
+      "loss": 1.134,
+      "step": 28920
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.1597589106332315e-05,
+      "loss": 1.1238,
+      "step": 28930
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.159461269439691e-05,
+      "loss": 1.141,
+      "step": 28940
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.1591636282461496e-05,
+      "loss": 1.13,
+      "step": 28950
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.158865987052608e-05,
+      "loss": 1.1321,
+      "step": 28960
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 3.158568345859067e-05,
+      "loss": 1.1343,
+      "step": 28970
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.158270704665526e-05,
+      "loss": 1.1032,
+      "step": 28980
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.157973063471985e-05,
+      "loss": 1.1169,
+      "step": 28990
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.157675422278444e-05,
+      "loss": 1.1137,
+      "step": 29000
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.1573777810849024e-05,
+      "loss": 1.1316,
+      "step": 29010
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.157080139891361e-05,
+      "loss": 1.1171,
+      "step": 29020
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.15678249869782e-05,
+      "loss": 1.1375,
+      "step": 29030
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.1564848575042785e-05,
+      "loss": 1.1128,
+      "step": 29040
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.156187216310738e-05,
+      "loss": 1.1189,
+      "step": 29050
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 3.1558895751171965e-05,
+      "loss": 1.1294,
+      "step": 29060
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.155591933923655e-05,
+      "loss": 1.1471,
+      "step": 29070
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.155294292730114e-05,
+      "loss": 1.1112,
+      "step": 29080
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.154996651536573e-05,
+      "loss": 1.1197,
+      "step": 29090
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.154699010343032e-05,
+      "loss": 1.1393,
+      "step": 29100
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.154401369149491e-05,
+      "loss": 1.1133,
+      "step": 29110
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.1541037279559494e-05,
+      "loss": 1.1222,
+      "step": 29120
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.153806086762408e-05,
+      "loss": 1.1113,
+      "step": 29130
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.153508445568867e-05,
+      "loss": 1.1236,
+      "step": 29140
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 3.153210804375326e-05,
+      "loss": 1.1384,
+      "step": 29150
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.152913163181785e-05,
+      "loss": 1.1239,
+      "step": 29160
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.1526155219882435e-05,
+      "loss": 1.1304,
+      "step": 29170
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.152317880794702e-05,
+      "loss": 1.1289,
+      "step": 29180
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.1520202396011615e-05,
+      "loss": 1.121,
+      "step": 29190
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.15172259840762e-05,
+      "loss": 1.1353,
+      "step": 29200
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.151424957214078e-05,
+      "loss": 1.1376,
+      "step": 29210
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.1511273160205376e-05,
+      "loss": 1.1226,
+      "step": 29220
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.150829674826996e-05,
+      "loss": 1.1338,
+      "step": 29230
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 3.150532033633455e-05,
+      "loss": 1.1265,
+      "step": 29240
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.150234392439914e-05,
+      "loss": 1.1243,
+      "step": 29250
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.149936751246373e-05,
+      "loss": 1.1345,
+      "step": 29260
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.149639110052832e-05,
+      "loss": 1.1246,
+      "step": 29270
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.1493414688592904e-05,
+      "loss": 1.1379,
+      "step": 29280
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.149043827665749e-05,
+      "loss": 1.1341,
+      "step": 29290
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.1487461864722085e-05,
+      "loss": 1.1309,
+      "step": 29300
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.1484485452786665e-05,
+      "loss": 1.1342,
+      "step": 29310
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.148150904085125e-05,
+      "loss": 1.1222,
+      "step": 29320
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.1478532628915845e-05,
+      "loss": 1.1231,
+      "step": 29330
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.147555621698043e-05,
+      "loss": 1.1441,
+      "step": 29340
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.147257980504502e-05,
+      "loss": 1.1206,
+      "step": 29350
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.1469603393109606e-05,
+      "loss": 1.1161,
+      "step": 29360
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.14666269811742e-05,
+      "loss": 1.1315,
+      "step": 29370
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.146365056923879e-05,
+      "loss": 1.1078,
+      "step": 29380
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.1460674157303374e-05,
+      "loss": 1.118,
+      "step": 29390
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.145769774536796e-05,
+      "loss": 1.1187,
+      "step": 29400
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.1454721333432554e-05,
+      "loss": 1.1127,
+      "step": 29410
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.1451744921497134e-05,
+      "loss": 1.1383,
+      "step": 29420
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.144876850956172e-05,
+      "loss": 1.1189,
+      "step": 29430
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.1445792097626315e-05,
+      "loss": 1.134,
+      "step": 29440
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.14428156856909e-05,
+      "loss": 1.1223,
+      "step": 29450
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.143983927375549e-05,
+      "loss": 1.1308,
+      "step": 29460
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.1436862861820076e-05,
+      "loss": 1.119,
+      "step": 29470
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.143388644988467e-05,
+      "loss": 1.1153,
+      "step": 29480
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.1430910037949256e-05,
+      "loss": 1.1558,
+      "step": 29490
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.142793362601384e-05,
+      "loss": 1.1243,
+      "step": 29500
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.142495721407843e-05,
+      "loss": 1.1338,
+      "step": 29510
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.142198080214302e-05,
+      "loss": 1.118,
+      "step": 29520
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.1419004390207604e-05,
+      "loss": 1.1217,
+      "step": 29530
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.14160279782722e-05,
+      "loss": 1.1374,
+      "step": 29540
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.1413051566336784e-05,
+      "loss": 1.1226,
+      "step": 29550
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.141007515440137e-05,
+      "loss": 1.1346,
+      "step": 29560
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.140709874246596e-05,
+      "loss": 1.1196,
+      "step": 29570
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.140412233053055e-05,
+      "loss": 1.1292,
+      "step": 29580
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.140114591859514e-05,
+      "loss": 1.1318,
+      "step": 29590
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.1398169506659726e-05,
+      "loss": 1.1225,
+      "step": 29600
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.139519309472431e-05,
+      "loss": 1.122,
+      "step": 29610
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.13922166827889e-05,
+      "loss": 1.1412,
+      "step": 29620
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.1389240270853486e-05,
+      "loss": 1.1219,
+      "step": 29630
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.138626385891807e-05,
+      "loss": 1.1357,
+      "step": 29640
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.138328744698267e-05,
+      "loss": 1.1313,
+      "step": 29650
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.1380311035047254e-05,
+      "loss": 1.1126,
+      "step": 29660
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.137733462311184e-05,
+      "loss": 1.1343,
+      "step": 29670
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.137435821117643e-05,
+      "loss": 1.1277,
+      "step": 29680
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.137138179924102e-05,
+      "loss": 1.1217,
+      "step": 29690
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.136840538730561e-05,
+      "loss": 1.1416,
+      "step": 29700
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.1365428975370195e-05,
+      "loss": 1.117,
+      "step": 29710
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.136245256343478e-05,
+      "loss": 1.1167,
+      "step": 29720
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.135947615149937e-05,
+      "loss": 1.1483,
+      "step": 29730
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.1356499739563956e-05,
+      "loss": 1.1145,
+      "step": 29740
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.135352332762854e-05,
+      "loss": 1.1219,
+      "step": 29750
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.1350546915693136e-05,
+      "loss": 1.1155,
+      "step": 29760
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.134757050375772e-05,
+      "loss": 1.1266,
+      "step": 29770
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.134459409182231e-05,
+      "loss": 1.1166,
+      "step": 29780
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.13416176798869e-05,
+      "loss": 1.1224,
+      "step": 29790
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.133864126795149e-05,
+      "loss": 1.1386,
+      "step": 29800
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.133566485601608e-05,
+      "loss": 1.1232,
+      "step": 29810
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.133268844408066e-05,
+      "loss": 1.126,
+      "step": 29820
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.132971203214525e-05,
+      "loss": 1.1311,
+      "step": 29830
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.132673562020984e-05,
+      "loss": 1.1323,
+      "step": 29840
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.1323759208274425e-05,
+      "loss": 1.1349,
+      "step": 29850
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.132078279633901e-05,
+      "loss": 1.1225,
+      "step": 29860
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 3.1317806384403606e-05,
+      "loss": 1.1259,
+      "step": 29870
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.131482997246819e-05,
+      "loss": 1.1296,
+      "step": 29880
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.131185356053278e-05,
+      "loss": 1.1408,
+      "step": 29890
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.1308877148597366e-05,
+      "loss": 1.1273,
+      "step": 29900
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.130590073666196e-05,
+      "loss": 1.1156,
+      "step": 29910
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.130292432472655e-05,
+      "loss": 1.131,
+      "step": 29920
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.1299947912791134e-05,
+      "loss": 1.1199,
+      "step": 29930
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.129697150085572e-05,
+      "loss": 1.1149,
+      "step": 29940
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.129399508892031e-05,
+      "loss": 1.1407,
+      "step": 29950
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 3.1291018676984895e-05,
+      "loss": 1.1223,
+      "step": 29960
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.128804226504949e-05,
+      "loss": 1.1176,
+      "step": 29970
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.1285065853114075e-05,
+      "loss": 1.1468,
+      "step": 29980
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.128208944117866e-05,
+      "loss": 1.1408,
+      "step": 29990
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.127911302924325e-05,
+      "loss": 1.1168,
+      "step": 30000
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.127613661730784e-05,
+      "loss": 1.1361,
+      "step": 30010
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.127316020537243e-05,
+      "loss": 1.1485,
+      "step": 30020
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.127018379343701e-05,
+      "loss": 1.1246,
+      "step": 30030
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.1267207381501603e-05,
+      "loss": 1.1108,
+      "step": 30040
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 3.126423096956619e-05,
+      "loss": 1.1206,
+      "step": 30050
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.126125455763078e-05,
+      "loss": 1.1317,
+      "step": 30060
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.1258278145695364e-05,
+      "loss": 1.1298,
+      "step": 30070
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.125530173375996e-05,
+      "loss": 1.1227,
+      "step": 30080
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.1252325321824545e-05,
+      "loss": 1.1315,
+      "step": 30090
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.124934890988913e-05,
+      "loss": 1.1173,
+      "step": 30100
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.124637249795372e-05,
+      "loss": 1.1185,
+      "step": 30110
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.124339608601831e-05,
+      "loss": 1.1294,
+      "step": 30120
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.124041967408289e-05,
+      "loss": 1.1248,
+      "step": 30130
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 3.123744326214748e-05,
+      "loss": 1.1358,
+      "step": 30140
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.123446685021207e-05,
+      "loss": 1.1086,
+      "step": 30150
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.123149043827666e-05,
+      "loss": 1.1317,
+      "step": 30160
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.122851402634125e-05,
+      "loss": 1.1362,
+      "step": 30170
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.1225537614405834e-05,
+      "loss": 1.1352,
+      "step": 30180
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.122256120247043e-05,
+      "loss": 1.1316,
+      "step": 30190
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.1219584790535014e-05,
+      "loss": 1.137,
+      "step": 30200
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.12166083785996e-05,
+      "loss": 1.1266,
+      "step": 30210
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.121363196666419e-05,
+      "loss": 1.1364,
+      "step": 30220
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.1210655554728775e-05,
+      "loss": 1.1388,
+      "step": 30230
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.120767914279336e-05,
+      "loss": 1.1174,
+      "step": 30240
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.120470273085795e-05,
+      "loss": 1.13,
+      "step": 30250
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.120172631892254e-05,
+      "loss": 1.1159,
+      "step": 30260
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.119874990698713e-05,
+      "loss": 1.1228,
+      "step": 30270
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 3.1195773495051716e-05,
+      "loss": 1.1107,
+      "step": 30280
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.11927970831163e-05,
+      "loss": 1.1429,
+      "step": 30290
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.11898206711809e-05,
+      "loss": 1.1291,
+      "step": 30300
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.1186844259245484e-05,
+      "loss": 1.1276,
+      "step": 30310
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.118386784731007e-05,
+      "loss": 1.1422,
+      "step": 30320
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.118089143537466e-05,
+      "loss": 1.1201,
+      "step": 30330
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.1177915023439244e-05,
+      "loss": 1.1284,
+      "step": 30340
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.117493861150383e-05,
+      "loss": 1.1268,
+      "step": 30350
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.1171962199568425e-05,
+      "loss": 1.1335,
+      "step": 30360
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.116898578763301e-05,
+      "loss": 1.1319,
+      "step": 30370
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.11660093756976e-05,
+      "loss": 1.1098,
+      "step": 30380
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.1163032963762186e-05,
+      "loss": 1.1225,
+      "step": 30390
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.116005655182678e-05,
+      "loss": 1.1228,
+      "step": 30400
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.1157080139891366e-05,
+      "loss": 1.1335,
+      "step": 30410
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.115410372795595e-05,
+      "loss": 1.1154,
+      "step": 30420
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.115112731602054e-05,
+      "loss": 1.1265,
+      "step": 30430
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.114815090408513e-05,
+      "loss": 1.1424,
+      "step": 30440
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.1145174492149714e-05,
+      "loss": 1.1253,
+      "step": 30450
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.11421980802143e-05,
+      "loss": 1.121,
+      "step": 30460
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.1139221668278894e-05,
+      "loss": 1.1141,
+      "step": 30470
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.113624525634348e-05,
+      "loss": 1.1222,
+      "step": 30480
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.113326884440807e-05,
+      "loss": 1.123,
+      "step": 30490
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.1130292432472655e-05,
+      "loss": 1.1235,
+      "step": 30500
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.112731602053725e-05,
+      "loss": 1.1258,
+      "step": 30510
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.1124339608601836e-05,
+      "loss": 1.1138,
+      "step": 30520
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.112136319666642e-05,
+      "loss": 1.1194,
+      "step": 30530
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.111838678473101e-05,
+      "loss": 1.1051,
+      "step": 30540
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.1115410372795596e-05,
+      "loss": 1.1307,
+      "step": 30550
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.111243396086018e-05,
+      "loss": 1.131,
+      "step": 30560
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.110945754892477e-05,
+      "loss": 1.1211,
+      "step": 30570
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.1106481136989364e-05,
+      "loss": 1.1225,
+      "step": 30580
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.110350472505395e-05,
+      "loss": 1.1164,
+      "step": 30590
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.110052831311854e-05,
+      "loss": 1.1258,
+      "step": 30600
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.1097551901183124e-05,
+      "loss": 1.1271,
+      "step": 30610
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.109457548924772e-05,
+      "loss": 1.1307,
+      "step": 30620
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.1091599077312305e-05,
+      "loss": 1.1249,
+      "step": 30630
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.108862266537689e-05,
+      "loss": 1.1296,
+      "step": 30640
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.108564625344148e-05,
+      "loss": 1.143,
+      "step": 30650
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.1082669841506066e-05,
+      "loss": 1.1116,
+      "step": 30660
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.107969342957065e-05,
+      "loss": 1.1189,
+      "step": 30670
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.107671701763524e-05,
+      "loss": 1.1131,
+      "step": 30680
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.107374060569983e-05,
+      "loss": 1.124,
+      "step": 30690
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.107076419376442e-05,
+      "loss": 1.1301,
+      "step": 30700
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.106778778182901e-05,
+      "loss": 1.1325,
+      "step": 30710
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.1064811369893594e-05,
+      "loss": 1.1236,
+      "step": 30720
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.106183495795819e-05,
+      "loss": 1.1327,
+      "step": 30730
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.1058858546022774e-05,
+      "loss": 1.1263,
+      "step": 30740
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.105588213408736e-05,
+      "loss": 1.1412,
+      "step": 30750
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.105290572215195e-05,
+      "loss": 1.1339,
+      "step": 30760
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.1049929310216535e-05,
+      "loss": 1.1069,
+      "step": 30770
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.104695289828112e-05,
+      "loss": 1.1239,
+      "step": 30780
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.1043976486345716e-05,
+      "loss": 1.1165,
+      "step": 30790
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.10410000744103e-05,
+      "loss": 1.1249,
+      "step": 30800
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.103802366247489e-05,
+      "loss": 1.1207,
+      "step": 30810
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.1035047250539476e-05,
+      "loss": 1.1202,
+      "step": 30820
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.103207083860407e-05,
+      "loss": 1.1311,
+      "step": 30830
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.102909442666866e-05,
+      "loss": 1.1298,
+      "step": 30840
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.102611801473324e-05,
+      "loss": 1.1296,
+      "step": 30850
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.102314160279783e-05,
+      "loss": 1.1194,
+      "step": 30860
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.102016519086242e-05,
+      "loss": 1.0913,
+      "step": 30870
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.1017188778927005e-05,
+      "loss": 1.1,
+      "step": 30880
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.101421236699159e-05,
+      "loss": 1.1321,
+      "step": 30890
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.1011235955056185e-05,
+      "loss": 1.1238,
+      "step": 30900
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.100825954312077e-05,
+      "loss": 1.107,
+      "step": 30910
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.100528313118536e-05,
+      "loss": 1.1029,
+      "step": 30920
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.1002306719249946e-05,
+      "loss": 1.1388,
+      "step": 30930
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.099933030731454e-05,
+      "loss": 1.1106,
+      "step": 30940
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 3.099635389537912e-05,
+      "loss": 1.1046,
+      "step": 30950
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.0993377483443707e-05,
+      "loss": 1.1213,
+      "step": 30960
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.09904010715083e-05,
+      "loss": 1.1162,
+      "step": 30970
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.098742465957289e-05,
+      "loss": 1.1194,
+      "step": 30980
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.0984448247637474e-05,
+      "loss": 1.1286,
+      "step": 30990
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.098147183570206e-05,
+      "loss": 1.1218,
+      "step": 31000
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.0978495423766655e-05,
+      "loss": 1.1222,
+      "step": 31010
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.097551901183124e-05,
+      "loss": 1.1201,
+      "step": 31020
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.097254259989583e-05,
+      "loss": 1.1404,
+      "step": 31030
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.0969566187960415e-05,
+      "loss": 1.1062,
+      "step": 31040
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.0966589776025e-05,
+      "loss": 1.1231,
+      "step": 31050
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.096361336408959e-05,
+      "loss": 1.1253,
+      "step": 31060
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.096063695215418e-05,
+      "loss": 1.1078,
+      "step": 31070
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.095766054021877e-05,
+      "loss": 1.12,
+      "step": 31080
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.0954684128283357e-05,
+      "loss": 1.1208,
+      "step": 31090
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.0951707716347943e-05,
+      "loss": 1.1257,
+      "step": 31100
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.094873130441253e-05,
+      "loss": 1.1172,
+      "step": 31110
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.0945754892477124e-05,
+      "loss": 1.1225,
+      "step": 31120
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.094277848054171e-05,
+      "loss": 1.1278,
+      "step": 31130
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.09398020686063e-05,
+      "loss": 1.113,
+      "step": 31140
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.0936825656670885e-05,
+      "loss": 1.1259,
+      "step": 31150
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.093384924473547e-05,
+      "loss": 1.1201,
+      "step": 31160
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.093087283280006e-05,
+      "loss": 1.1239,
+      "step": 31170
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.092789642086465e-05,
+      "loss": 1.1115,
+      "step": 31180
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.092492000892924e-05,
+      "loss": 1.1187,
+      "step": 31190
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.0921943596993826e-05,
+      "loss": 1.1173,
+      "step": 31200
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.091896718505841e-05,
+      "loss": 1.1202,
+      "step": 31210
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.0915990773123007e-05,
+      "loss": 1.1244,
+      "step": 31220
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.0913014361187593e-05,
+      "loss": 1.1265,
+      "step": 31230
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.091003794925218e-05,
+      "loss": 1.1279,
+      "step": 31240
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.090706153731677e-05,
+      "loss": 1.1164,
+      "step": 31250
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.0904085125381354e-05,
+      "loss": 1.1329,
+      "step": 31260
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.090110871344594e-05,
+      "loss": 1.1204,
+      "step": 31270
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.089813230151053e-05,
+      "loss": 1.0962,
+      "step": 31280
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.089515588957512e-05,
+      "loss": 1.1207,
+      "step": 31290
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.089217947763971e-05,
+      "loss": 1.1346,
+      "step": 31300
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.0889203065704295e-05,
+      "loss": 1.1132,
+      "step": 31310
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.088622665376888e-05,
+      "loss": 1.1248,
+      "step": 31320
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.0883250241833476e-05,
+      "loss": 1.1148,
+      "step": 31330
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.088027382989806e-05,
+      "loss": 1.1305,
+      "step": 31340
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.087729741796265e-05,
+      "loss": 1.133,
+      "step": 31350
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.087432100602724e-05,
+      "loss": 1.1272,
+      "step": 31360
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.0871344594091824e-05,
+      "loss": 1.1,
+      "step": 31370
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.086836818215641e-05,
+      "loss": 1.1141,
+      "step": 31380
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.0865391770221e-05,
+      "loss": 1.1087,
+      "step": 31390
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.086241535828559e-05,
+      "loss": 1.0915,
+      "step": 31400
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.085943894635018e-05,
+      "loss": 1.1273,
+      "step": 31410
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.0856462534414765e-05,
+      "loss": 1.1311,
+      "step": 31420
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.085348612247935e-05,
+      "loss": 1.1239,
+      "step": 31430
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.0850509710543945e-05,
+      "loss": 1.0923,
+      "step": 31440
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.084753329860853e-05,
+      "loss": 1.1282,
+      "step": 31450
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.084455688667312e-05,
+      "loss": 1.1237,
+      "step": 31460
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.0841580474737706e-05,
+      "loss": 1.1136,
+      "step": 31470
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.083860406280229e-05,
+      "loss": 1.1216,
+      "step": 31480
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.083562765086688e-05,
+      "loss": 1.1267,
+      "step": 31490
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.0832651238931474e-05,
+      "loss": 1.1189,
+      "step": 31500
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.082967482699606e-05,
+      "loss": 1.1188,
+      "step": 31510
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.082669841506065e-05,
+      "loss": 1.1155,
+      "step": 31520
+    },
+    {
+      "epoch": 7.0,
+      "eval_cer": 4.764957264957265,
+      "eval_loss": 1.1188170909881592,
+      "eval_runtime": 5.4262,
+      "eval_samples_per_second": 1.843,
+      "eval_steps_per_second": 0.184,
+      "eval_wer": 1.0129870129870129,
+      "step": 31521
+    }
+  ],
+  "max_steps": 135090,
+  "num_train_epochs": 30,
+  "total_flos": 1.1077976360516271e+21,
+  "trial_name": null,
+  "trial_params": null
+}