Upload 4 files

rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4292d1113b4a8d3acc9212b18e9fa8b8bd873c8a52928d290e0302ec7df29163
-size 14575
+oid sha256:8389f21e71ce92f43a547972b89c1c26b0f480b5298694cb63fb2754cff0298a
+size 14511

scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:eb75cd671247b41bbd34848e4df7927814e754c9cb28ec0aa9edae59ac06cf69
+oid sha256:267eadca69671bc0460ea479ba8c13a4276356a16351a73e44520e14439a1b48
 size 627

trainer_state.json

@@ -1,763 +1,2434 @@
 {
-  "best_metric": 349.6192030463756,
-  "best_model_checkpoint": "./whisper-small-th/checkpoint-1000",
-  "epoch": 2.2440917272493515,
-  "global_step": 3000,
+  "best_metric": 455.00475996192034,
+  "best_model_checkpoint": "./whisper-small-th/checkpoint-10000",
+  "epoch": 3.7397157816005984,
+  "global_step": 10000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.02,
-      "learning_rate": 5.000000000000001e-07,
-      "loss": 1.4088,
+      "epoch": 0.01,
+      "learning_rate": 4.4e-07,
+      "loss": 1.2638,
       "step": 25
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.0000000000000002e-06,
-      "loss": 1.1763,
+      "epoch": 0.02,
+      "learning_rate": 9.400000000000001e-07,
+      "loss": 1.1302,
       "step": 50
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.5e-06,
-      "loss": 0.7903,
+      "epoch": 0.03,
+      "learning_rate": 1.44e-06,
+      "loss": 0.8499,
       "step": 75
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 2.0000000000000003e-06,
-      "loss": 0.5779,
+      "epoch": 0.04,
+      "learning_rate": 1.94e-06,
+      "loss": 0.5874,
       "step": 100
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 2.5e-06,
-      "loss": 0.5041,
+      "epoch": 0.05,
+      "learning_rate": 2.4400000000000004e-06,
+      "loss": 0.5495,
       "step": 125
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 3e-06,
-      "loss": 0.4802,
+      "epoch": 0.06,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 0.5288,
       "step": 150
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 3.48e-06,
-      "loss": 0.4499,
+      "epoch": 0.07,
+      "learning_rate": 3.44e-06,
+      "loss": 0.4926,
       "step": 175
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 3.980000000000001e-06,
-      "loss": 0.447,
+      "epoch": 0.07,
+      "learning_rate": 3.94e-06,
+      "loss": 0.4848,
       "step": 200
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.48e-06,
-      "loss": 0.4707,
+      "epoch": 0.08,
+      "learning_rate": 4.440000000000001e-06,
+      "loss": 0.4442,
       "step": 225
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.980000000000001e-06,
-      "loss": 0.4222,
+      "epoch": 0.09,
+      "learning_rate": 4.94e-06,
+      "loss": 0.447,
       "step": 250
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 5.480000000000001e-06,
-      "loss": 0.4262,
+      "epoch": 0.1,
+      "learning_rate": 5.4400000000000004e-06,
+      "loss": 0.417,
       "step": 275
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 5.98e-06,
-      "loss": 0.3833,
+      "epoch": 0.11,
+      "learning_rate": 5.94e-06,
+      "loss": 0.4122,
       "step": 300
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 6.480000000000001e-06,
-      "loss": 0.3818,
+      "epoch": 0.12,
+      "learning_rate": 6.440000000000001e-06,
+      "loss": 0.3987,
       "step": 325
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 6.98e-06,
-      "loss": 0.3956,
+      "epoch": 0.13,
+      "learning_rate": 6.9400000000000005e-06,
+      "loss": 0.3986,
       "step": 350
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 7.48e-06,
-      "loss": 0.4027,
+      "epoch": 0.14,
+      "learning_rate": 7.440000000000001e-06,
+      "loss": 0.4054,
       "step": 375
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 7.980000000000002e-06,
-      "loss": 0.3553,
+      "epoch": 0.15,
+      "learning_rate": 7.94e-06,
+      "loss": 0.3771,
       "step": 400
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 8.48e-06,
-      "loss": 0.36,
+      "epoch": 0.16,
+      "learning_rate": 8.44e-06,
+      "loss": 0.4053,
       "step": 425
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 8.98e-06,
-      "loss": 0.346,
+      "epoch": 0.17,
+      "learning_rate": 8.94e-06,
+      "loss": 0.3854,
       "step": 450
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 9.48e-06,
-      "loss": 0.3413,
+      "epoch": 0.18,
+      "learning_rate": 9.440000000000001e-06,
+      "loss": 0.3699,
       "step": 475
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 9.980000000000001e-06,
-      "loss": 0.3354,
+      "epoch": 0.19,
+      "learning_rate": 9.940000000000001e-06,
+      "loss": 0.355,
       "step": 500
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 9.946666666666667e-06,
-      "loss": 0.3542,
+      "epoch": 0.2,
+      "learning_rate": 9.976842105263158e-06,
+      "loss": 0.3734,
       "step": 525
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 9.891111111111113e-06,
-      "loss": 0.3199,
+      "epoch": 0.21,
+      "learning_rate": 9.950526315789475e-06,
+      "loss": 0.3522,
       "step": 550
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 9.835555555555556e-06,
-      "loss": 0.3121,
+      "epoch": 0.22,
+      "learning_rate": 9.92421052631579e-06,
+      "loss": 0.3296,
       "step": 575
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 9.780000000000001e-06,
-      "loss": 0.3131,
+      "epoch": 0.22,
+      "learning_rate": 9.897894736842107e-06,
+      "loss": 0.3258,
       "step": 600
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 9.724444444444445e-06,
-      "loss": 0.3073,
+      "epoch": 0.23,
+      "learning_rate": 9.871578947368422e-06,
+      "loss": 0.3292,
       "step": 625
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 9.66888888888889e-06,
-      "loss": 0.312,
+      "epoch": 0.24,
+      "learning_rate": 9.845263157894738e-06,
+      "loss": 0.3251,
       "step": 650
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 9.613333333333335e-06,
-      "loss": 0.3004,
+      "epoch": 0.25,
+      "learning_rate": 9.818947368421053e-06,
+      "loss": 0.3431,
       "step": 675
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 9.557777777777777e-06,
-      "loss": 0.2856,
+      "epoch": 0.26,
+      "learning_rate": 9.79263157894737e-06,
+      "loss": 0.3267,
       "step": 700
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 9.502222222222223e-06,
-      "loss": 0.2822,
+      "epoch": 0.27,
+      "learning_rate": 9.766315789473685e-06,
+      "loss": 0.336,
       "step": 725
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 9.446666666666667e-06,
-      "loss": 0.2953,
+      "epoch": 0.28,
+      "learning_rate": 9.74e-06,
+      "loss": 0.3626,
       "step": 750
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 9.391111111111111e-06,
-      "loss": 0.2755,
+      "epoch": 0.29,
+      "learning_rate": 9.713684210526317e-06,
+      "loss": 0.3152,
       "step": 775
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 9.335555555555557e-06,
-      "loss": 0.2797,
+      "epoch": 0.3,
+      "learning_rate": 9.687368421052632e-06,
+      "loss": 0.2871,
       "step": 800
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 9.280000000000001e-06,
-      "loss": 0.2648,
+      "epoch": 0.31,
+      "learning_rate": 9.661052631578948e-06,
+      "loss": 0.3152,
       "step": 825
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 9.224444444444445e-06,
-      "loss": 0.2957,
+      "epoch": 0.32,
+      "learning_rate": 9.634736842105265e-06,
+      "loss": 0.3014,
       "step": 850
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 9.168888888888889e-06,
-      "loss": 0.2802,
+      "epoch": 0.33,
+      "learning_rate": 9.60842105263158e-06,
+      "loss": 0.2967,
       "step": 875
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 9.113333333333335e-06,
-      "loss": 0.2613,
+      "epoch": 0.34,
+      "learning_rate": 9.582105263157897e-06,
+      "loss": 0.3026,
       "step": 900
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 9.057777777777779e-06,
-      "loss": 0.2729,
+      "epoch": 0.35,
+      "learning_rate": 9.555789473684211e-06,
+      "loss": 0.2893,
       "step": 925
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 9.002222222222223e-06,
-      "loss": 0.2506,
+      "epoch": 0.36,
+      "learning_rate": 9.529473684210528e-06,
+      "loss": 0.2983,
       "step": 950
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 8.946666666666669e-06,
-      "loss": 0.2542,
+      "epoch": 0.36,
+      "learning_rate": 9.503157894736843e-06,
+      "loss": 0.2903,
       "step": 975
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 8.891111111111111e-06,
-      "loss": 0.261,
-      "step": 1000
-    },
-    {
-      "epoch": 0.75,
-      "eval_loss": 0.22164225578308105,
-      "eval_runtime": 6851.3771,
-      "eval_samples_per_second": 1.595,
-      "eval_steps_per_second": 1.595,
-      "eval_wer": 349.6192030463756,
+      "epoch": 0.37,
+      "learning_rate": 9.476842105263158e-06,
+      "loss": 0.2734,
       "step": 1000
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 8.835555555555557e-06,
-      "loss": 0.262,
+      "epoch": 0.38,
+      "learning_rate": 9.450526315789475e-06,
+      "loss": 0.2947,
       "step": 1025
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 8.78e-06,
-      "loss": 0.2451,
+      "epoch": 0.39,
+      "learning_rate": 9.42421052631579e-06,
+      "loss": 0.2876,
       "step": 1050
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 8.724444444444445e-06,
-      "loss": 0.2607,
+      "epoch": 0.4,
+      "learning_rate": 9.397894736842106e-06,
+      "loss": 0.275,
       "step": 1075
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 8.66888888888889e-06,
-      "loss": 0.2613,
+      "epoch": 0.41,
+      "learning_rate": 9.371578947368421e-06,
+      "loss": 0.2818,
       "step": 1100
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 8.613333333333333e-06,
-      "loss": 0.2346,
+      "epoch": 0.42,
+      "learning_rate": 9.345263157894738e-06,
+      "loss": 0.2714,
       "step": 1125
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 8.557777777777778e-06,
-      "loss": 0.2353,
+      "epoch": 0.43,
+      "learning_rate": 9.318947368421053e-06,
+      "loss": 0.2648,
       "step": 1150
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 8.502222222222223e-06,
-      "loss": 0.2456,
+      "epoch": 0.44,
+      "learning_rate": 9.292631578947368e-06,
+      "loss": 0.2845,
       "step": 1175
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 8.446666666666668e-06,
-      "loss": 0.2266,
+      "epoch": 0.45,
+      "learning_rate": 9.266315789473685e-06,
+      "loss": 0.2606,
       "step": 1200
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 8.391111111111112e-06,
-      "loss": 0.2259,
+      "epoch": 0.46,
+      "learning_rate": 9.240000000000001e-06,
+      "loss": 0.2757,
       "step": 1225
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 8.335555555555556e-06,
-      "loss": 0.2314,
+      "epoch": 0.47,
+      "learning_rate": 9.213684210526316e-06,
+      "loss": 0.2595,
       "step": 1250
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 8.28e-06,
-      "loss": 0.2279,
+      "epoch": 0.48,
+      "learning_rate": 9.187368421052633e-06,
+      "loss": 0.2589,
       "step": 1275
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 8.224444444444444e-06,
-      "loss": 0.2484,
+      "epoch": 0.49,
+      "learning_rate": 9.161052631578948e-06,
+      "loss": 0.2851,
       "step": 1300
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 8.16888888888889e-06,
-      "loss": 0.2277,
+      "epoch": 0.5,
+      "learning_rate": 9.134736842105265e-06,
+      "loss": 0.2528,
       "step": 1325
     },
     {
-      "epoch": 1.01,
-      "learning_rate": 8.113333333333334e-06,
-      "loss": 0.2045,
+      "epoch": 0.5,
+      "learning_rate": 9.10842105263158e-06,
+      "loss": 0.2596,
       "step": 1350
     },
     {
-      "epoch": 1.03,
-      "learning_rate": 8.057777777777778e-06,
-      "loss": 0.1655,
+      "epoch": 0.51,
+      "learning_rate": 9.082105263157896e-06,
+      "loss": 0.2529,
       "step": 1375
     },
     {
-      "epoch": 1.05,
-      "learning_rate": 8.002222222222222e-06,
-      "loss": 0.1579,
+      "epoch": 0.52,
+      "learning_rate": 9.055789473684211e-06,
+      "loss": 0.255,
       "step": 1400
     },
     {
-      "epoch": 1.07,
-      "learning_rate": 7.946666666666666e-06,
-      "loss": 0.1664,
+      "epoch": 0.53,
+      "learning_rate": 9.029473684210526e-06,
+      "loss": 0.2448,
       "step": 1425
     },
     {
-      "epoch": 1.08,
-      "learning_rate": 7.891111111111112e-06,
-      "loss": 0.1616,
+      "epoch": 0.54,
+      "learning_rate": 9.003157894736843e-06,
+      "loss": 0.2463,
       "step": 1450
     },
     {
-      "epoch": 1.1,
-      "learning_rate": 7.835555555555556e-06,
-      "loss": 0.1649,
+      "epoch": 0.55,
+      "learning_rate": 8.976842105263158e-06,
+      "loss": 0.2353,
       "step": 1475
     },
     {
-      "epoch": 1.12,
-      "learning_rate": 7.78e-06,
-      "loss": 0.1629,
+      "epoch": 0.56,
+      "learning_rate": 8.950526315789474e-06,
+      "loss": 0.276,
       "step": 1500
     },
     {
-      "epoch": 1.14,
-      "learning_rate": 7.724444444444446e-06,
-      "loss": 0.1567,
+      "epoch": 0.57,
+      "learning_rate": 8.92421052631579e-06,
+      "loss": 0.2342,
       "step": 1525
     },
     {
-      "epoch": 1.16,
-      "learning_rate": 7.66888888888889e-06,
-      "loss": 0.1532,
+      "epoch": 0.58,
+      "learning_rate": 8.897894736842106e-06,
+      "loss": 0.2423,
       "step": 1550
     },
     {
-      "epoch": 1.18,
-      "learning_rate": 7.613333333333334e-06,
-      "loss": 0.1574,
+      "epoch": 0.59,
+      "learning_rate": 8.871578947368423e-06,
+      "loss": 0.2332,
       "step": 1575
     },
     {
-      "epoch": 1.2,
-      "learning_rate": 7.557777777777779e-06,
-      "loss": 0.1597,
+      "epoch": 0.6,
+      "learning_rate": 8.845263157894738e-06,
+      "loss": 0.2464,
       "step": 1600
     },
     {
-      "epoch": 1.22,
-      "learning_rate": 7.502222222222223e-06,
-      "loss": 0.1824,
+      "epoch": 0.61,
+      "learning_rate": 8.818947368421054e-06,
+      "loss": 0.2404,
       "step": 1625
     },
     {
-      "epoch": 1.23,
-      "learning_rate": 7.446666666666668e-06,
-      "loss": 0.157,
+      "epoch": 0.62,
+      "learning_rate": 8.79263157894737e-06,
+      "loss": 0.2332,
       "step": 1650
     },
     {
-      "epoch": 1.25,
-      "learning_rate": 7.3911111111111125e-06,
-      "loss": 0.1496,
+      "epoch": 0.63,
+      "learning_rate": 8.766315789473684e-06,
+      "loss": 0.266,
       "step": 1675
     },
     {
-      "epoch": 1.27,
-      "learning_rate": 7.335555555555556e-06,
-      "loss": 0.1515,
+      "epoch": 0.64,
+      "learning_rate": 8.740000000000001e-06,
+      "loss": 0.2578,
       "step": 1700
     },
     {
-      "epoch": 1.29,
-      "learning_rate": 7.280000000000001e-06,
-      "loss": 0.1485,
+      "epoch": 0.65,
+      "learning_rate": 8.713684210526316e-06,
+      "loss": 0.2342,
       "step": 1725
     },
     {
-      "epoch": 1.31,
-      "learning_rate": 7.224444444444445e-06,
-      "loss": 0.1443,
+      "epoch": 0.65,
+      "learning_rate": 8.687368421052633e-06,
+      "loss": 0.2404,
       "step": 1750
     },
     {
-      "epoch": 1.33,
-      "learning_rate": 7.1688888888888895e-06,
-      "loss": 0.1466,
+      "epoch": 0.66,
+      "learning_rate": 8.661052631578948e-06,
+      "loss": 0.2142,
       "step": 1775
     },
     {
-      "epoch": 1.35,
-      "learning_rate": 7.113333333333334e-06,
-      "loss": 0.1611,
+      "epoch": 0.67,
+      "learning_rate": 8.634736842105264e-06,
+      "loss": 0.2407,
       "step": 1800
     },
     {
-      "epoch": 1.37,
-      "learning_rate": 7.057777777777778e-06,
-      "loss": 0.1537,
+      "epoch": 0.68,
+      "learning_rate": 8.60842105263158e-06,
+      "loss": 0.2247,
       "step": 1825
     },
     {
-      "epoch": 1.38,
-      "learning_rate": 7.0022222222222225e-06,
-      "loss": 0.1515,
+      "epoch": 0.69,
+      "learning_rate": 8.582105263157894e-06,
+      "loss": 0.2416,
       "step": 1850
     },
     {
-      "epoch": 1.4,
-      "learning_rate": 6.946666666666667e-06,
-      "loss": 0.1612,
+      "epoch": 0.7,
+      "learning_rate": 8.555789473684213e-06,
+      "loss": 0.2256,
       "step": 1875
     },
     {
-      "epoch": 1.42,
-      "learning_rate": 6.891111111111111e-06,
-      "loss": 0.1484,
+      "epoch": 0.71,
+      "learning_rate": 8.529473684210528e-06,
+      "loss": 0.2279,
       "step": 1900
     },
     {
-      "epoch": 1.44,
-      "learning_rate": 6.835555555555556e-06,
-      "loss": 0.151,
+      "epoch": 0.72,
+      "learning_rate": 8.503157894736843e-06,
+      "loss": 0.2352,
       "step": 1925
     },
     {
-      "epoch": 1.46,
-      "learning_rate": 6.780000000000001e-06,
-      "loss": 0.1476,
+      "epoch": 0.73,
+      "learning_rate": 8.47684210526316e-06,
+      "loss": 0.2055,
       "step": 1950
     },
     {
-      "epoch": 1.48,
-      "learning_rate": 6.724444444444444e-06,
-      "loss": 0.151,
+      "epoch": 0.74,
+      "learning_rate": 8.450526315789474e-06,
+      "loss": 0.2436,
       "step": 1975
     },
     {
-      "epoch": 1.5,
-      "learning_rate": 6.668888888888889e-06,
-      "loss": 0.1471,
-      "step": 2000
-    },
-    {
-      "epoch": 1.5,
-      "eval_loss": 0.18018271028995514,
-      "eval_runtime": 6949.9088,
-      "eval_samples_per_second": 1.573,
-      "eval_steps_per_second": 1.573,
-      "eval_wer": 364.84428124575004,
+      "epoch": 0.75,
+      "learning_rate": 8.42421052631579e-06,
+      "loss": 0.2192,
       "step": 2000
     },
     {
-      "epoch": 1.51,
-      "learning_rate": 6.613333333333334e-06,
-      "loss": 0.1381,
+      "epoch": 0.76,
+      "learning_rate": 8.397894736842106e-06,
+      "loss": 0.2153,
       "step": 2025
     },
     {
-      "epoch": 1.53,
-      "learning_rate": 6.557777777777778e-06,
-      "loss": 0.1562,
+      "epoch": 0.77,
+      "learning_rate": 8.371578947368422e-06,
+      "loss": 0.2288,
       "step": 2050
     },
     {
-      "epoch": 1.55,
-      "learning_rate": 6.502222222222223e-06,
-      "loss": 0.1565,
+      "epoch": 0.78,
+      "learning_rate": 8.345263157894737e-06,
+      "loss": 0.2113,
       "step": 2075
     },
     {
-      "epoch": 1.57,
-      "learning_rate": 6.446666666666668e-06,
-      "loss": 0.1698,
+      "epoch": 0.79,
+      "learning_rate": 8.318947368421052e-06,
+      "loss": 0.22,
       "step": 2100
     },
     {
-      "epoch": 1.59,
-      "learning_rate": 6.391111111111111e-06,
-      "loss": 0.1423,
+      "epoch": 0.79,
+      "learning_rate": 8.292631578947369e-06,
+      "loss": 0.2353,
       "step": 2125
     },
     {
-      "epoch": 1.61,
-      "learning_rate": 6.335555555555556e-06,
-      "loss": 0.1409,
+      "epoch": 0.8,
+      "learning_rate": 8.266315789473684e-06,
+      "loss": 0.2166,
       "step": 2150
     },
     {
-      "epoch": 1.63,
-      "learning_rate": 6.280000000000001e-06,
-      "loss": 0.1637,
+      "epoch": 0.81,
+      "learning_rate": 8.24e-06,
+      "loss": 0.2308,
       "step": 2175
     },
     {
-      "epoch": 1.65,
-      "learning_rate": 6.224444444444445e-06,
-      "loss": 0.1445,
+      "epoch": 0.82,
+      "learning_rate": 8.213684210526316e-06,
+      "loss": 0.2223,
       "step": 2200
     },
     {
-      "epoch": 1.66,
-      "learning_rate": 6.16888888888889e-06,
-      "loss": 0.1404,
+      "epoch": 0.83,
+      "learning_rate": 8.187368421052632e-06,
+      "loss": 0.2062,
       "step": 2225
     },
     {
-      "epoch": 1.68,
-      "learning_rate": 6.113333333333333e-06,
-      "loss": 0.1558,
+      "epoch": 0.84,
+      "learning_rate": 8.161052631578949e-06,
+      "loss": 0.2152,
       "step": 2250
     },
     {
-      "epoch": 1.7,
-      "learning_rate": 6.057777777777778e-06,
-      "loss": 0.1454,
+      "epoch": 0.85,
+      "learning_rate": 8.134736842105264e-06,
+      "loss": 0.2102,
       "step": 2275
     },
     {
-      "epoch": 1.72,
-      "learning_rate": 6.002222222222223e-06,
-      "loss": 0.1409,
+      "epoch": 0.86,
+      "learning_rate": 8.10842105263158e-06,
+      "loss": 0.2109,
       "step": 2300
     },
     {
-      "epoch": 1.74,
-      "learning_rate": 5.946666666666668e-06,
-      "loss": 0.143,
+      "epoch": 0.87,
+      "learning_rate": 8.082105263157896e-06,
+      "loss": 0.2088,
       "step": 2325
     },
     {
-      "epoch": 1.76,
-      "learning_rate": 5.891111111111112e-06,
-      "loss": 0.1403,
+      "epoch": 0.88,
+      "learning_rate": 8.05578947368421e-06,
+      "loss": 0.2251,
       "step": 2350
     },
     {
-      "epoch": 1.78,
-      "learning_rate": 5.8355555555555565e-06,
-      "loss": 0.1523,
+      "epoch": 0.89,
+      "learning_rate": 8.029473684210527e-06,
+      "loss": 0.1999,
       "step": 2375
     },
     {
-      "epoch": 1.8,
-      "learning_rate": 5.78e-06,
-      "loss": 0.1308,
+      "epoch": 0.9,
+      "learning_rate": 8.003157894736842e-06,
+      "loss": 0.2096,
       "step": 2400
     },
     {
-      "epoch": 1.81,
-      "learning_rate": 5.724444444444445e-06,
-      "loss": 0.1508,
+      "epoch": 0.91,
+      "learning_rate": 7.976842105263159e-06,
+      "loss": 0.2063,
       "step": 2425
     },
     {
-      "epoch": 1.83,
-      "learning_rate": 5.6688888888888895e-06,
-      "loss": 0.1478,
+      "epoch": 0.92,
+      "learning_rate": 7.950526315789474e-06,
+      "loss": 0.1951,
       "step": 2450
     },
     {
-      "epoch": 1.85,
-      "learning_rate": 5.613333333333334e-06,
-      "loss": 0.1517,
+      "epoch": 0.93,
+      "learning_rate": 7.92421052631579e-06,
+      "loss": 0.2071,
       "step": 2475
     },
     {
-      "epoch": 1.87,
-      "learning_rate": 5.557777777777778e-06,
-      "loss": 0.136,
+      "epoch": 0.93,
+      "learning_rate": 7.897894736842106e-06,
+      "loss": 0.1938,
       "step": 2500
     },
     {
-      "epoch": 1.89,
-      "learning_rate": 5.5022222222222224e-06,
-      "loss": 0.1361,
+      "epoch": 0.94,
+      "learning_rate": 7.87157894736842e-06,
+      "loss": 0.1833,
       "step": 2525
     },
     {
-      "epoch": 1.91,
-      "learning_rate": 5.4466666666666665e-06,
-      "loss": 0.1419,
+      "epoch": 0.95,
+      "learning_rate": 7.845263157894739e-06,
+      "loss": 0.2094,
       "step": 2550
     },
     {
-      "epoch": 1.93,
-      "learning_rate": 5.391111111111111e-06,
-      "loss": 0.146,
+      "epoch": 0.96,
+      "learning_rate": 7.818947368421054e-06,
+      "loss": 0.2129,
       "step": 2575
     },
     {
-      "epoch": 1.94,
-      "learning_rate": 5.335555555555556e-06,
-      "loss": 0.1456,
+      "epoch": 0.97,
+      "learning_rate": 7.792631578947369e-06,
+      "loss": 0.2211,
       "step": 2600
     },
     {
-      "epoch": 1.96,
-      "learning_rate": 5.28e-06,
-      "loss": 0.1383,
+      "epoch": 0.98,
+      "learning_rate": 7.766315789473685e-06,
+      "loss": 0.1998,
       "step": 2625
     },
     {
-      "epoch": 1.98,
-      "learning_rate": 5.224444444444445e-06,
-      "loss": 0.142,
+      "epoch": 0.99,
+      "learning_rate": 7.74e-06,
+      "loss": 0.1936,
       "step": 2650
     },
     {
-      "epoch": 2.0,
-      "learning_rate": 5.168888888888889e-06,
-      "loss": 0.1353,
+      "epoch": 1.0,
+      "learning_rate": 7.713684210526317e-06,
+      "loss": 0.216,
       "step": 2675
     },
     {
-      "epoch": 2.02,
-      "learning_rate": 5.113333333333333e-06,
-      "loss": 0.0836,
+      "epoch": 1.01,
+      "learning_rate": 7.687368421052632e-06,
+      "loss": 0.1361,
       "step": 2700
     },
     {
-      "epoch": 2.04,
-      "learning_rate": 5.057777777777778e-06,
-      "loss": 0.081,
+      "epoch": 1.02,
+      "learning_rate": 7.661052631578949e-06,
+      "loss": 0.1317,
       "step": 2725
     },
     {
-      "epoch": 2.06,
-      "learning_rate": 5.002222222222223e-06,
-      "loss": 0.0914,
+      "epoch": 1.03,
+      "learning_rate": 7.634736842105264e-06,
+      "loss": 0.1479,
       "step": 2750
     },
     {
-      "epoch": 2.08,
-      "learning_rate": 4.946666666666667e-06,
-      "loss": 0.0806,
+      "epoch": 1.04,
+      "learning_rate": 7.6084210526315795e-06,
+      "loss": 0.1223,
       "step": 2775
     },
     {
-      "epoch": 2.09,
-      "learning_rate": 4.891111111111111e-06,
-      "loss": 0.083,
+      "epoch": 1.05,
+      "learning_rate": 7.582105263157895e-06,
+      "loss": 0.136,
       "step": 2800
     },
     {
-      "epoch": 2.11,
-      "learning_rate": 4.835555555555556e-06,
-      "loss": 0.078,
+      "epoch": 1.06,
+      "learning_rate": 7.555789473684211e-06,
+      "loss": 0.1326,
       "step": 2825
     },
     {
-      "epoch": 2.13,
-      "learning_rate": 4.78e-06,
-      "loss": 0.0799,
+      "epoch": 1.07,
+      "learning_rate": 7.529473684210526e-06,
+      "loss": 0.1389,
       "step": 2850
     },
     {
-      "epoch": 2.15,
-      "learning_rate": 4.724444444444445e-06,
-      "loss": 0.0864,
+      "epoch": 1.08,
+      "learning_rate": 7.503157894736842e-06,
+      "loss": 0.1281,
       "step": 2875
     },
     {
-      "epoch": 2.17,
-      "learning_rate": 4.66888888888889e-06,
-      "loss": 0.078,
+      "epoch": 1.08,
+      "learning_rate": 7.476842105263159e-06,
+      "loss": 0.1304,
       "step": 2900
     },
     {
-      "epoch": 2.19,
-      "learning_rate": 4.613333333333334e-06,
-      "loss": 0.0863,
+      "epoch": 1.09,
+      "learning_rate": 7.4505263157894744e-06,
+      "loss": 0.1369,
       "step": 2925
     },
     {
-      "epoch": 2.21,
-      "learning_rate": 4.557777777777778e-06,
-      "loss": 0.0794,
+      "epoch": 1.1,
+      "learning_rate": 7.42421052631579e-06,
+      "loss": 0.1298,
       "step": 2950
     },
     {
-      "epoch": 2.23,
-      "learning_rate": 4.502222222222223e-06,
-      "loss": 0.0789,
+      "epoch": 1.11,
+      "learning_rate": 7.397894736842106e-06,
+      "loss": 0.1381,
       "step": 2975
     },
     {
-      "epoch": 2.24,
-      "learning_rate": 4.446666666666667e-06,
-      "loss": 0.093,
+      "epoch": 1.12,
+      "learning_rate": 7.371578947368422e-06,
+      "loss": 0.1295,
       "step": 3000
     },
     {
-      "epoch": 2.24,
-      "eval_loss": 0.16714586317539215,
-      "eval_runtime": 6790.9362,
-      "eval_samples_per_second": 1.609,
-      "eval_steps_per_second": 1.609,
-      "eval_wer": 360.4719162246702,
-      "step": 3000
+      "epoch": 1.13,
+      "learning_rate": 7.345263157894738e-06,
+      "loss": 0.1207,
+      "step": 3025
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.3189473684210535e-06,
+      "loss": 0.1275,
+      "step": 3050
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.292631578947369e-06,
+      "loss": 0.1276,
+      "step": 3075
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.266315789473684e-06,
+      "loss": 0.1249,
+      "step": 3100
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.24e-06,
+      "loss": 0.1247,
+      "step": 3125
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.213684210526316e-06,
+      "loss": 0.1284,
+      "step": 3150
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.187368421052632e-06,
+      "loss": 0.1253,
+      "step": 3175
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.161052631578948e-06,
+      "loss": 0.134,
+      "step": 3200
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.134736842105264e-06,
+      "loss": 0.1452,
+      "step": 3225
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.10842105263158e-06,
+      "loss": 0.1579,
+      "step": 3250
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.082105263157896e-06,
+      "loss": 0.1282,
+      "step": 3275
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.055789473684212e-06,
+      "loss": 0.1314,
+      "step": 3300
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.0294736842105275e-06,
+      "loss": 0.115,
+      "step": 3325
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.0031578947368425e-06,
+      "loss": 0.137,
+      "step": 3350
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.976842105263158e-06,
+      "loss": 0.1145,
+      "step": 3375
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.950526315789474e-06,
+      "loss": 0.1378,
+      "step": 3400
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.92421052631579e-06,
+      "loss": 0.1218,
+      "step": 3425
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.897894736842106e-06,
+      "loss": 0.1218,
+      "step": 3450
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.871578947368422e-06,
+      "loss": 0.118,
+      "step": 3475
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.845263157894737e-06,
+      "loss": 0.1192,
+      "step": 3500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.818947368421052e-06,
+      "loss": 0.1162,
+      "step": 3525
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.792631578947368e-06,
+      "loss": 0.1225,
+      "step": 3550
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.766315789473685e-06,
+      "loss": 0.1302,
+      "step": 3575
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.740000000000001e-06,
+      "loss": 0.1425,
+      "step": 3600
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.7136842105263165e-06,
+      "loss": 0.1166,
+      "step": 3625
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.687368421052632e-06,
+      "loss": 0.1307,
+      "step": 3650
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.661052631578948e-06,
+      "loss": 0.1366,
+      "step": 3675
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.634736842105264e-06,
+      "loss": 0.1209,
+      "step": 3700
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.60842105263158e-06,
+      "loss": 0.1359,
+      "step": 3725
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.582105263157896e-06,
+      "loss": 0.1288,
+      "step": 3750
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.5557894736842106e-06,
+      "loss": 0.1205,
+      "step": 3775
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.529473684210526e-06,
+      "loss": 0.122,
+      "step": 3800
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.503157894736842e-06,
+      "loss": 0.1242,
+      "step": 3825
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.476842105263158e-06,
+      "loss": 0.1156,
+      "step": 3850
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.450526315789474e-06,
+      "loss": 0.1233,
+      "step": 3875
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.42421052631579e-06,
+      "loss": 0.1171,
+      "step": 3900
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.397894736842106e-06,
+      "loss": 0.1294,
+      "step": 3925
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.371578947368422e-06,
+      "loss": 0.1217,
+      "step": 3950
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.345263157894738e-06,
+      "loss": 0.1305,
+      "step": 3975
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.318947368421054e-06,
+      "loss": 0.1166,
+      "step": 4000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.292631578947369e-06,
+      "loss": 0.1159,
+      "step": 4025
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.266315789473685e-06,
+      "loss": 0.1193,
+      "step": 4050
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.24e-06,
+      "loss": 0.1286,
+      "step": 4075
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.213684210526316e-06,
+      "loss": 0.1237,
+      "step": 4100
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.187368421052632e-06,
+      "loss": 0.1299,
+      "step": 4125
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.161052631578948e-06,
+      "loss": 0.1286,
+      "step": 4150
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.134736842105264e-06,
+      "loss": 0.1458,
+      "step": 4175
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.108421052631579e-06,
+      "loss": 0.133,
+      "step": 4200
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.0821052631578945e-06,
+      "loss": 0.1091,
+      "step": 4225
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.055789473684212e-06,
+      "loss": 0.1205,
+      "step": 4250
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.029473684210527e-06,
+      "loss": 0.1151,
+      "step": 4275
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.003157894736843e-06,
+      "loss": 0.119,
+      "step": 4300
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.976842105263159e-06,
+      "loss": 0.1374,
+      "step": 4325
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.950526315789474e-06,
+      "loss": 0.1307,
+      "step": 4350
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.92421052631579e-06,
+      "loss": 0.1207,
+      "step": 4375
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.897894736842106e-06,
+      "loss": 0.1104,
+      "step": 4400
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.871578947368422e-06,
+      "loss": 0.1074,
+      "step": 4425
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.845263157894737e-06,
+      "loss": 0.1256,
+      "step": 4450
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.818947368421053e-06,
+      "loss": 0.1345,
+      "step": 4475
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.7926315789473685e-06,
+      "loss": 0.1171,
+      "step": 4500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.766315789473684e-06,
+      "loss": 0.1166,
+      "step": 4525
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.74e-06,
+      "loss": 0.1242,
+      "step": 4550
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.713684210526316e-06,
+      "loss": 0.1148,
+      "step": 4575
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.687368421052633e-06,
+      "loss": 0.12,
+      "step": 4600
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.661052631578948e-06,
+      "loss": 0.1255,
+      "step": 4625
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.634736842105264e-06,
+      "loss": 0.1083,
+      "step": 4650
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.60842105263158e-06,
+      "loss": 0.1199,
+      "step": 4675
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.582105263157895e-06,
+      "loss": 0.1183,
+      "step": 4700
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.555789473684211e-06,
+      "loss": 0.1213,
+      "step": 4725
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.529473684210527e-06,
+      "loss": 0.132,
+      "step": 4750
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.5031578947368425e-06,
+      "loss": 0.112,
+      "step": 4775
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.476842105263158e-06,
+      "loss": 0.107,
+      "step": 4800
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.450526315789474e-06,
+      "loss": 0.1277,
+      "step": 4825
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.42421052631579e-06,
+      "loss": 0.1206,
+      "step": 4850
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.397894736842105e-06,
+      "loss": 0.1144,
+      "step": 4875
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.371578947368421e-06,
+      "loss": 0.1301,
+      "step": 4900
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.345263157894738e-06,
+      "loss": 0.1204,
+      "step": 4925
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.318947368421053e-06,
+      "loss": 0.1301,
+      "step": 4950
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.292631578947369e-06,
+      "loss": 0.108,
+      "step": 4975
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.266315789473685e-06,
+      "loss": 0.1186,
+      "step": 5000
+    },
+    {
+      "epoch": 1.87,
+      "eval_loss": 0.15749630331993103,
+      "eval_runtime": 5122.7824,
+      "eval_samples_per_second": 2.134,
+      "eval_steps_per_second": 0.267,
+      "eval_wer": 511.77750577995374,
+      "step": 5000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.240000000000001e-06,
+      "loss": 0.1077,
+      "step": 5025
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.2136842105263165e-06,
+      "loss": 0.1183,
+      "step": 5050
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.187368421052632e-06,
+      "loss": 0.122,
+      "step": 5075
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.161052631578948e-06,
+      "loss": 0.1202,
+      "step": 5100
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.134736842105263e-06,
+      "loss": 0.127,
+      "step": 5125
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.108421052631579e-06,
+      "loss": 0.1197,
+      "step": 5150
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.082105263157895e-06,
+      "loss": 0.1142,
+      "step": 5175
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.0557894736842105e-06,
+      "loss": 0.1361,
+      "step": 5200
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.030526315789475e-06,
+      "loss": 0.1096,
+      "step": 5225
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.0042105263157906e-06,
+      "loss": 0.1158,
+      "step": 5250
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.9778947368421055e-06,
+      "loss": 0.1218,
+      "step": 5275
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.951578947368421e-06,
+      "loss": 0.1121,
+      "step": 5300
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.925263157894737e-06,
+      "loss": 0.1224,
+      "step": 5325
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.898947368421053e-06,
+      "loss": 0.1031,
+      "step": 5350
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.872631578947369e-06,
+      "loss": 0.0677,
+      "step": 5375
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.846315789473685e-06,
+      "loss": 0.0618,
+      "step": 5400
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.8200000000000004e-06,
+      "loss": 0.06,
+      "step": 5425
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.793684210526316e-06,
+      "loss": 0.0526,
+      "step": 5450
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.767368421052632e-06,
+      "loss": 0.0642,
+      "step": 5475
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.741052631578948e-06,
+      "loss": 0.0702,
+      "step": 5500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.714736842105264e-06,
+      "loss": 0.0583,
+      "step": 5525
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.6884210526315795e-06,
+      "loss": 0.0608,
+      "step": 5550
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.662105263157895e-06,
+      "loss": 0.0633,
+      "step": 5575
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.63578947368421e-06,
+      "loss": 0.0572,
+      "step": 5600
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.609473684210526e-06,
+      "loss": 0.0644,
+      "step": 5625
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.583157894736843e-06,
+      "loss": 0.0573,
+      "step": 5650
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.556842105263159e-06,
+      "loss": 0.0568,
+      "step": 5675
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.5305263157894744e-06,
+      "loss": 0.0675,
+      "step": 5700
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.504210526315789e-06,
+      "loss": 0.0729,
+      "step": 5725
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.477894736842105e-06,
+      "loss": 0.0606,
+      "step": 5750
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.451578947368421e-06,
+      "loss": 0.0535,
+      "step": 5775
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.425263157894737e-06,
+      "loss": 0.061,
+      "step": 5800
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.3989473684210535e-06,
+      "loss": 0.0597,
+      "step": 5825
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.3726315789473685e-06,
+      "loss": 0.0621,
+      "step": 5850
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.346315789473684e-06,
+      "loss": 0.0603,
+      "step": 5875
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.32e-06,
+      "loss": 0.0523,
+      "step": 5900
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.293684210526316e-06,
+      "loss": 0.0629,
+      "step": 5925
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.267368421052632e-06,
+      "loss": 0.0578,
+      "step": 5950
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.241052631578948e-06,
+      "loss": 0.0644,
+      "step": 5975
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.214736842105263e-06,
+      "loss": 0.0685,
+      "step": 6000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.188421052631579e-06,
+      "loss": 0.0671,
+      "step": 6025
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.162105263157895e-06,
+      "loss": 0.0534,
+      "step": 6050
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.135789473684211e-06,
+      "loss": 0.0619,
+      "step": 6075
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.109473684210527e-06,
+      "loss": 0.0612,
+      "step": 6100
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.0831578947368425e-06,
+      "loss": 0.0648,
+      "step": 6125
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.056842105263158e-06,
+      "loss": 0.0572,
+      "step": 6150
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.030526315789474e-06,
+      "loss": 0.0556,
+      "step": 6175
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.00421052631579e-06,
+      "loss": 0.0592,
+      "step": 6200
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.977894736842106e-06,
+      "loss": 0.0592,
+      "step": 6225
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.951578947368422e-06,
+      "loss": 0.0612,
+      "step": 6250
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.9252631578947366e-06,
+      "loss": 0.055,
+      "step": 6275
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.898947368421052e-06,
+      "loss": 0.059,
+      "step": 6300
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.872631578947369e-06,
+      "loss": 0.0643,
+      "step": 6325
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.846315789473685e-06,
+      "loss": 0.0639,
+      "step": 6350
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.820000000000001e-06,
+      "loss": 0.0472,
+      "step": 6375
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.793684210526316e-06,
+      "loss": 0.0661,
+      "step": 6400
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.767368421052632e-06,
+      "loss": 0.0604,
+      "step": 6425
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.7410526315789473e-06,
+      "loss": 0.0654,
+      "step": 6450
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.714736842105263e-06,
+      "loss": 0.0639,
+      "step": 6475
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.6884210526315794e-06,
+      "loss": 0.0586,
+      "step": 6500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.662105263157895e-06,
+      "loss": 0.0566,
+      "step": 6525
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.635789473684211e-06,
+      "loss": 0.056,
+      "step": 6550
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6094736842105264e-06,
+      "loss": 0.0592,
+      "step": 6575
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5831578947368422e-06,
+      "loss": 0.0585,
+      "step": 6600
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.556842105263158e-06,
+      "loss": 0.0549,
+      "step": 6625
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5305263157894743e-06,
+      "loss": 0.0658,
+      "step": 6650
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.50421052631579e-06,
+      "loss": 0.062,
+      "step": 6675
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.4778947368421055e-06,
+      "loss": 0.0572,
+      "step": 6700
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.4515789473684213e-06,
+      "loss": 0.0592,
+      "step": 6725
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.425263157894737e-06,
+      "loss": 0.0732,
+      "step": 6750
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.398947368421053e-06,
+      "loss": 0.0605,
+      "step": 6775
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.3726315789473683e-06,
+      "loss": 0.065,
+      "step": 6800
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.3463157894736846e-06,
+      "loss": 0.0698,
+      "step": 6825
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.3200000000000004e-06,
+      "loss": 0.0638,
+      "step": 6850
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.2936842105263162e-06,
+      "loss": 0.0588,
+      "step": 6875
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.267368421052632e-06,
+      "loss": 0.0659,
+      "step": 6900
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.2410526315789474e-06,
+      "loss": 0.0635,
+      "step": 6925
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.2147368421052633e-06,
+      "loss": 0.0614,
+      "step": 6950
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.188421052631579e-06,
+      "loss": 0.062,
+      "step": 6975
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.1621052631578953e-06,
+      "loss": 0.0657,
+      "step": 7000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.135789473684211e-06,
+      "loss": 0.0565,
+      "step": 7025
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.1094736842105265e-06,
+      "loss": 0.0571,
+      "step": 7050
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.0831578947368423e-06,
+      "loss": 0.0526,
+      "step": 7075
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.056842105263158e-06,
+      "loss": 0.051,
+      "step": 7100
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.0305263157894736e-06,
+      "loss": 0.0562,
+      "step": 7125
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.0042105263157894e-06,
+      "loss": 0.0685,
+      "step": 7150
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.9778947368421056e-06,
+      "loss": 0.0615,
+      "step": 7175
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.9515789473684214e-06,
+      "loss": 0.0615,
+      "step": 7200
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.9252631578947373e-06,
+      "loss": 0.0665,
+      "step": 7225
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.8989473684210526e-06,
+      "loss": 0.06,
+      "step": 7250
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.8726315789473685e-06,
+      "loss": 0.0534,
+      "step": 7275
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.8463157894736843e-06,
+      "loss": 0.0576,
+      "step": 7300
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.82e-06,
+      "loss": 0.0543,
+      "step": 7325
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.7936842105263163e-06,
+      "loss": 0.064,
+      "step": 7350
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.7673684210526317e-06,
+      "loss": 0.0601,
+      "step": 7375
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.7410526315789476e-06,
+      "loss": 0.0516,
+      "step": 7400
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7147368421052634e-06,
+      "loss": 0.0587,
+      "step": 7425
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.688421052631579e-06,
+      "loss": 0.0566,
+      "step": 7450
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.6621052631578946e-06,
+      "loss": 0.0542,
+      "step": 7475
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.635789473684211e-06,
+      "loss": 0.0576,
+      "step": 7500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.6094736842105267e-06,
+      "loss": 0.0526,
+      "step": 7525
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.5831578947368425e-06,
+      "loss": 0.0574,
+      "step": 7550
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.5568421052631583e-06,
+      "loss": 0.049,
+      "step": 7575
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.5305263157894737e-06,
+      "loss": 0.0624,
+      "step": 7600
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5042105263157895e-06,
+      "loss": 0.0589,
+      "step": 7625
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.4778947368421053e-06,
+      "loss": 0.062,
+      "step": 7650
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.451578947368421e-06,
+      "loss": 0.0629,
+      "step": 7675
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.425263157894737e-06,
+      "loss": 0.0626,
+      "step": 7700
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.3989473684210528e-06,
+      "loss": 0.0569,
+      "step": 7725
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.3726315789473686e-06,
+      "loss": 0.0636,
+      "step": 7750
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.3463157894736844e-06,
+      "loss": 0.0519,
+      "step": 7775
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.3200000000000002e-06,
+      "loss": 0.0525,
+      "step": 7800
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.293684210526316e-06,
+      "loss": 0.0576,
+      "step": 7825
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.267368421052632e-06,
+      "loss": 0.063,
+      "step": 7850
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.2410526315789473e-06,
+      "loss": 0.0494,
+      "step": 7875
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.2147368421052635e-06,
+      "loss": 0.0614,
+      "step": 7900
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.188421052631579e-06,
+      "loss": 0.0579,
+      "step": 7925
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.1621052631578947e-06,
+      "loss": 0.0525,
+      "step": 7950
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.135789473684211e-06,
+      "loss": 0.0496,
+      "step": 7975
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.1094736842105264e-06,
+      "loss": 0.052,
+      "step": 8000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.083157894736842e-06,
+      "loss": 0.0587,
+      "step": 8025
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.056842105263158e-06,
+      "loss": 0.0285,
+      "step": 8050
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.030526315789474e-06,
+      "loss": 0.0319,
+      "step": 8075
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.0042105263157896e-06,
+      "loss": 0.0286,
+      "step": 8100
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9778947368421055e-06,
+      "loss": 0.0301,
+      "step": 8125
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9515789473684213e-06,
+      "loss": 0.0299,
+      "step": 8150
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.925263157894737e-06,
+      "loss": 0.0294,
+      "step": 8175
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8989473684210527e-06,
+      "loss": 0.0307,
+      "step": 8200
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.8726315789473687e-06,
+      "loss": 0.0337,
+      "step": 8225
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.8463157894736843e-06,
+      "loss": 0.0298,
+      "step": 8250
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.8200000000000002e-06,
+      "loss": 0.024,
+      "step": 8275
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.7936842105263158e-06,
+      "loss": 0.028,
+      "step": 8300
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.7673684210526318e-06,
+      "loss": 0.0269,
+      "step": 8325
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.7410526315789474e-06,
+      "loss": 0.0283,
+      "step": 8350
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.7147368421052632e-06,
+      "loss": 0.0283,
+      "step": 8375
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6884210526315792e-06,
+      "loss": 0.0293,
+      "step": 8400
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.6621052631578948e-06,
+      "loss": 0.0343,
+      "step": 8425
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.6357894736842107e-06,
+      "loss": 0.0305,
+      "step": 8450
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.6094736842105265e-06,
+      "loss": 0.0248,
+      "step": 8475
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5831578947368423e-06,
+      "loss": 0.0307,
+      "step": 8500
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.556842105263158e-06,
+      "loss": 0.0263,
+      "step": 8525
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5305263157894737e-06,
+      "loss": 0.0314,
+      "step": 8550
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5042105263157898e-06,
+      "loss": 0.0298,
+      "step": 8575
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4778947368421054e-06,
+      "loss": 0.0239,
+      "step": 8600
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4515789473684212e-06,
+      "loss": 0.0329,
+      "step": 8625
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.425263157894737e-06,
+      "loss": 0.0296,
+      "step": 8650
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3989473684210528e-06,
+      "loss": 0.0298,
+      "step": 8675
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3726315789473684e-06,
+      "loss": 0.0247,
+      "step": 8700
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3463157894736842e-06,
+      "loss": 0.0284,
+      "step": 8725
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.32e-06,
+      "loss": 0.0322,
+      "step": 8750
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.2936842105263159e-06,
+      "loss": 0.0285,
+      "step": 8775
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.2673684210526315e-06,
+      "loss": 0.0287,
+      "step": 8800
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.2410526315789475e-06,
+      "loss": 0.0317,
+      "step": 8825
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.2147368421052633e-06,
+      "loss": 0.0278,
+      "step": 8850
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.188421052631579e-06,
+      "loss": 0.029,
+      "step": 8875
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.1621052631578948e-06,
+      "loss": 0.0285,
+      "step": 8900
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.1357894736842106e-06,
+      "loss": 0.0257,
+      "step": 8925
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1094736842105264e-06,
+      "loss": 0.026,
+      "step": 8950
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.0831578947368422e-06,
+      "loss": 0.0303,
+      "step": 8975
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.0568421052631578e-06,
+      "loss": 0.029,
+      "step": 9000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.0305263157894739e-06,
+      "loss": 0.0282,
+      "step": 9025
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.0042105263157897e-06,
+      "loss": 0.029,
+      "step": 9050
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 9.778947368421053e-07,
+      "loss": 0.0252,
+      "step": 9075
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 9.515789473684212e-07,
+      "loss": 0.0273,
+      "step": 9100
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 9.252631578947368e-07,
+      "loss": 0.0316,
+      "step": 9125
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.989473684210527e-07,
+      "loss": 0.0276,
+      "step": 9150
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.726315789473686e-07,
+      "loss": 0.0287,
+      "step": 9175
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.463157894736843e-07,
+      "loss": 0.028,
+      "step": 9200
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.200000000000001e-07,
+      "loss": 0.0247,
+      "step": 9225
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.936842105263158e-07,
+      "loss": 0.0288,
+      "step": 9250
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.673684210526316e-07,
+      "loss": 0.026,
+      "step": 9275
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.410526315789475e-07,
+      "loss": 0.0304,
+      "step": 9300
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.147368421052632e-07,
+      "loss": 0.0283,
+      "step": 9325
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.884210526315791e-07,
+      "loss": 0.0264,
+      "step": 9350
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.621052631578948e-07,
+      "loss": 0.033,
+      "step": 9375
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.357894736842106e-07,
+      "loss": 0.0276,
+      "step": 9400
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.094736842105263e-07,
+      "loss": 0.0285,
+      "step": 9425
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.831578947368421e-07,
+      "loss": 0.0279,
+      "step": 9450
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.56842105263158e-07,
+      "loss": 0.0273,
+      "step": 9475
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.305263157894737e-07,
+      "loss": 0.0225,
+      "step": 9500
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.042105263157895e-07,
+      "loss": 0.0253,
+      "step": 9525
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.778947368421053e-07,
+      "loss": 0.0306,
+      "step": 9550
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.5157894736842107e-07,
+      "loss": 0.0294,
+      "step": 9575
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.2526315789473684e-07,
+      "loss": 0.0277,
+      "step": 9600
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.9894736842105266e-07,
+      "loss": 0.031,
+      "step": 9625
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.726315789473685e-07,
+      "loss": 0.031,
+      "step": 9650
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.4631578947368424e-07,
+      "loss": 0.0271,
+      "step": 9675
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.2e-07,
+      "loss": 0.024,
+      "step": 9700
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.936842105263158e-07,
+      "loss": 0.0283,
+      "step": 9725
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.6736842105263164e-07,
+      "loss": 0.0241,
+      "step": 9750
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.410526315789474e-07,
+      "loss": 0.0273,
+      "step": 9775
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.1473684210526317e-07,
+      "loss": 0.0315,
+      "step": 9800
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.8842105263157897e-07,
+      "loss": 0.0286,
+      "step": 9825
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.6210526315789476e-07,
+      "loss": 0.0274,
+      "step": 9850
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.3578947368421055e-07,
+      "loss": 0.0307,
+      "step": 9875
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.0947368421052632e-07,
+      "loss": 0.0296,
+      "step": 9900
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.315789473684211e-08,
+      "loss": 0.0277,
+      "step": 9925
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.68421052631579e-08,
+      "loss": 0.0261,
+      "step": 9950
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.0526315789473686e-08,
+      "loss": 0.025,
+      "step": 9975
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.210526315789474e-09,
+      "loss": 0.0237,
+      "step": 10000
+    },
+    {
+      "epoch": 3.74,
+      "eval_loss": 0.1606837958097458,
+      "eval_runtime": 4871.4962,
+      "eval_samples_per_second": 2.244,
+      "eval_steps_per_second": 0.281,
+      "eval_wer": 455.00475996192034,
+      "step": 10000
     }
   ],
-  "max_steps": 5000,
+  "max_steps": 10000,
   "num_train_epochs": 4,
-  "total_flos": 2.770419843072e+19,
+  "total_flos": 4.61693352701952e+19,
   "trial_name": null,
   "trial_params": null
 }