End of training

all_results.json

@@ -1,14 +1,14 @@
 {
-    "epoch": 2.0,
-    "eval_loss": Infinity,
-    "eval_runtime": 1256.035,
-    "eval_samples": 26618,
-    "eval_samples_per_second": 21.192,
-    "eval_steps_per_second": 0.662,
-    "eval_wer": 1.0,
-    "train_loss": 3.6355768978691203,
-    "train_runtime": 71670.0084,
-    "train_samples": 239774,
-    "train_samples_per_second": 6.691,
-    "train_steps_per_second": 0.052
+    "epoch": 49.99,
+    "eval_loss": 0.14973030984401703,
+    "eval_runtime": 191.6982,
+    "eval_samples": 4843,
+    "eval_samples_per_second": 25.264,
+    "eval_steps_per_second": 0.793,
+    "eval_wer": 0.1260857223956855,
+    "train_loss": 1.3462627654023223,
+    "train_runtime": 35973.7,
+    "train_samples": 11690,
+    "train_samples_per_second": 16.248,
+    "train_steps_per_second": 0.126
 }

eval_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 2.0,
-    "eval_loss": Infinity,
-    "eval_runtime": 1256.035,
-    "eval_samples": 26618,
-    "eval_samples_per_second": 21.192,
-    "eval_steps_per_second": 0.662,
-    "eval_wer": 1.0
+    "epoch": 49.99,
+    "eval_loss": 0.14973030984401703,
+    "eval_runtime": 191.6982,
+    "eval_samples": 4843,
+    "eval_samples_per_second": 25.264,
+    "eval_steps_per_second": 0.793,
+    "eval_wer": 0.1260857223956855
 }

train_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 2.0,
-    "train_loss": 3.6355768978691203,
-    "train_runtime": 71670.0084,
-    "train_samples": 239774,
-    "train_samples_per_second": 6.691,
-    "train_steps_per_second": 0.052
+    "epoch": 49.99,
+    "train_loss": 1.3462627654023223,
+    "train_runtime": 35973.7,
+    "train_samples": 11690,
+    "train_samples_per_second": 16.248,
+    "train_steps_per_second": 0.126
 }

trainer_state.json

@@ -1,1480 +1,1792 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 1.9998665421059656,
-  "global_step": 3746,
+  "epoch": 49.994535519125684,
+  "global_step": 4550,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.01,
-      "learning_rate": 0.0002,
-      "loss": 10.6012,
+      "epoch": 0.22,
+      "learning_rate": 2.5e-06,
+      "loss": 3.5867,
       "step": 20
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 0.0004,
-      "loss": 3.651,
+      "epoch": 0.44,
+      "learning_rate": 5e-06,
+      "loss": 3.5457,
       "step": 40
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 0.00059,
-      "loss": 3.3911,
+      "epoch": 0.66,
+      "learning_rate": 7.5e-06,
+      "loss": 3.4513,
       "step": 60
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 0.0007491827839825661,
-      "loss": 3.0481,
+      "epoch": 0.87,
+      "learning_rate": 1e-05,
+      "loss": 3.3432,
       "step": 80
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 0.0007453010078997548,
-      "loss": 3.4039,
+      "epoch": 1.1,
+      "learning_rate": 1.25e-05,
+      "loss": 3.3533,
       "step": 100
     },
     {
-      "epoch": 0.05,
-      "eval_loss": Infinity,
-      "eval_runtime": 1260.1628,
-      "eval_samples_per_second": 21.123,
-      "eval_steps_per_second": 0.66,
+      "epoch": 1.1,
+      "eval_loss": 3.2806732654571533,
+      "eval_runtime": 190.4728,
+      "eval_samples_per_second": 25.426,
+      "eval_steps_per_second": 0.798,
       "eval_wer": 1.0,
       "step": 100
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 0.0007412149278125851,
-      "loss": 3.0553,
+      "epoch": 1.32,
+      "learning_rate": 1.5e-05,
+      "loss": 3.2217,
       "step": 120
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 0.0007371288477254155,
-      "loss": 3.2191,
+      "epoch": 1.54,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 3.1765,
       "step": 140
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 0.0007332470716426042,
-      "loss": 3.4094,
+      "epoch": 1.75,
+      "learning_rate": 2e-05,
+      "loss": 3.1408,
       "step": 160
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 0.0007291609915554345,
-      "loss": 3.0607,
+      "epoch": 1.97,
+      "learning_rate": 2.2499999999999998e-05,
+      "loss": 3.1165,
       "step": 180
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 0.0007252792154726233,
-      "loss": 3.4396,
+      "epoch": 2.2,
+      "learning_rate": 2.5e-05,
+      "loss": 3.1709,
       "step": 200
     },
     {
-      "epoch": 0.11,
-      "eval_loss": Infinity,
-      "eval_runtime": 1235.1686,
-      "eval_samples_per_second": 21.55,
-      "eval_steps_per_second": 0.674,
+      "epoch": 2.2,
+      "eval_loss": 3.1325438022613525,
+      "eval_runtime": 192.4978,
+      "eval_samples_per_second": 25.159,
+      "eval_steps_per_second": 0.79,
       "eval_wer": 1.0,
       "step": 200
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 0.0007211931353854536,
-      "loss": 3.1198,
+      "epoch": 2.42,
+      "learning_rate": 2.75e-05,
+      "loss": 3.079,
       "step": 220
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 0.0007171070552982838,
-      "loss": 3.1801,
+      "epoch": 2.63,
+      "learning_rate": 3e-05,
+      "loss": 3.0677,
       "step": 240
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 0.0007132252792154726,
-      "loss": 3.3954,
+      "epoch": 2.85,
+      "learning_rate": 3.2500000000000004e-05,
+      "loss": 3.0656,
       "step": 260
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 0.000709139199128303,
-      "loss": 3.0566,
+      "epoch": 3.08,
+      "learning_rate": 3.5000000000000004e-05,
+      "loss": 3.1463,
       "step": 280
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 0.0007052574230454917,
-      "loss": 3.483,
+      "epoch": 3.3,
+      "learning_rate": 3.75e-05,
+      "loss": 3.0573,
       "step": 300
     },
     {
-      "epoch": 0.16,
-      "eval_loss": Infinity,
-      "eval_runtime": 1262.2253,
-      "eval_samples_per_second": 21.088,
-      "eval_steps_per_second": 0.659,
+      "epoch": 3.3,
+      "eval_loss": 3.0614514350891113,
+      "eval_runtime": 194.36,
+      "eval_samples_per_second": 24.918,
+      "eval_steps_per_second": 0.782,
       "eval_wer": 1.0,
       "step": 300
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 0.000701171342958322,
-      "loss": 3.1616,
+      "epoch": 3.51,
+      "learning_rate": 4e-05,
+      "loss": 3.0511,
       "step": 320
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 0.0006970852628711522,
-      "loss": 3.2017,
+      "epoch": 3.73,
+      "learning_rate": 4.25e-05,
+      "loss": 3.0358,
       "step": 340
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 0.0006932034867883411,
-      "loss": 3.4474,
+      "epoch": 3.95,
+      "learning_rate": 4.4999999999999996e-05,
+      "loss": 3.0416,
       "step": 360
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 0.0006891174067011714,
-      "loss": 3.0656,
+      "epoch": 4.17,
+      "learning_rate": 4.75e-05,
+      "loss": 3.1053,
       "step": 380
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 0.0006852356306183601,
-      "loss": 3.5014,
+      "epoch": 4.39,
+      "learning_rate": 5e-05,
+      "loss": 3.0314,
       "step": 400
     },
     {
-      "epoch": 0.21,
-      "eval_loss": Infinity,
-      "eval_runtime": 1255.7751,
-      "eval_samples_per_second": 21.196,
-      "eval_steps_per_second": 0.663,
+      "epoch": 4.39,
+      "eval_loss": 3.0990231037139893,
+      "eval_runtime": 198.3688,
+      "eval_samples_per_second": 24.414,
+      "eval_steps_per_second": 0.766,
       "eval_wer": 1.0,
       "step": 400
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 0.0006811495505311904,
-      "loss": 3.1757,
+      "epoch": 4.61,
+      "learning_rate": 5.25e-05,
+      "loss": 3.0309,
       "step": 420
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 0.0006770634704440206,
-      "loss": 3.2038,
+      "epoch": 4.83,
+      "learning_rate": 5.5e-05,
+      "loss": 3.0259,
       "step": 440
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 0.0006731816943612095,
-      "loss": 3.4072,
+      "epoch": 5.05,
+      "learning_rate": 5.75e-05,
+      "loss": 3.0998,
       "step": 460
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 0.0006690956142740398,
-      "loss": 3.0498,
+      "epoch": 5.27,
+      "learning_rate": 6e-05,
+      "loss": 3.0152,
       "step": 480
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 0.00066500953418687,
-      "loss": 3.331,
+      "epoch": 5.49,
+      "learning_rate": 6.25e-05,
+      "loss": 3.0129,
       "step": 500
     },
     {
-      "epoch": 0.27,
-      "eval_loss": Infinity,
-      "eval_runtime": 1239.2922,
-      "eval_samples_per_second": 21.478,
-      "eval_steps_per_second": 0.671,
+      "epoch": 5.49,
+      "eval_loss": 3.039973497390747,
+      "eval_runtime": 190.8567,
+      "eval_samples_per_second": 25.375,
+      "eval_steps_per_second": 0.796,
       "eval_wer": 1.0,
       "step": 500
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 0.0006609234540997004,
-      "loss": 3.0833,
+      "epoch": 5.71,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 3.0088,
       "step": 520
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 0.0006568373740125306,
-      "loss": 3.1989,
+      "epoch": 5.93,
+      "learning_rate": 6.75e-05,
+      "loss": 3.0051,
       "step": 540
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 0.0006531599019340779,
-      "loss": 3.5001,
+      "epoch": 6.15,
+      "learning_rate": 7.000000000000001e-05,
+      "loss": 3.073,
       "step": 560
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 0.0006490738218469082,
-      "loss": 3.0685,
+      "epoch": 6.37,
+      "learning_rate": 7.25e-05,
+      "loss": 3.0031,
       "step": 580
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 0.000645192045764097,
-      "loss": 3.4809,
+      "epoch": 6.59,
+      "learning_rate": 7.5e-05,
+      "loss": 2.9964,
       "step": 600
     },
     {
-      "epoch": 0.32,
-      "eval_loss": Infinity,
-      "eval_runtime": 1232.3119,
-      "eval_samples_per_second": 21.6,
-      "eval_steps_per_second": 0.675,
+      "epoch": 6.59,
+      "eval_loss": 2.998962640762329,
+      "eval_runtime": 193.5213,
+      "eval_samples_per_second": 25.026,
+      "eval_steps_per_second": 0.785,
       "eval_wer": 1.0,
       "step": 600
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 0.0006411059656769273,
-      "loss": 3.1877,
+      "epoch": 6.81,
+      "learning_rate": 7.75e-05,
+      "loss": 2.9921,
       "step": 620
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 0.0006370198855897576,
-      "loss": 3.2365,
+      "epoch": 7.03,
+      "learning_rate": 8e-05,
+      "loss": 3.0665,
       "step": 640
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 0.0006333424135113049,
-      "loss": 3.67,
+      "epoch": 7.25,
+      "learning_rate": 8.25e-05,
+      "loss": 2.9826,
       "step": 660
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 0.0006292563334241351,
-      "loss": 3.086,
+      "epoch": 7.47,
+      "learning_rate": 8.5e-05,
+      "loss": 2.9689,
       "step": 680
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 0.0006253745573413239,
-      "loss": 3.4678,
+      "epoch": 7.69,
+      "learning_rate": 8.75e-05,
+      "loss": 2.9602,
       "step": 700
     },
     {
-      "epoch": 0.37,
-      "eval_loss": Infinity,
-      "eval_runtime": 1236.8236,
-      "eval_samples_per_second": 21.521,
-      "eval_steps_per_second": 0.673,
+      "epoch": 7.69,
+      "eval_loss": 2.9620397090911865,
+      "eval_runtime": 193.5851,
+      "eval_samples_per_second": 25.017,
+      "eval_steps_per_second": 0.785,
       "eval_wer": 1.0,
       "step": 700
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 0.0006212884772541542,
-      "loss": 3.2344,
+      "epoch": 7.91,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 2.9639,
       "step": 720
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 0.0006172023971669845,
-      "loss": 3.2507,
+      "epoch": 8.13,
+      "learning_rate": 9.25e-05,
+      "loss": 3.0215,
       "step": 740
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 0.0006131163170798147,
-      "loss": 3.4283,
+      "epoch": 8.35,
+      "learning_rate": 9.5e-05,
+      "loss": 2.9454,
       "step": 760
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 0.0006090302369926451,
-      "loss": 3.0802,
+      "epoch": 8.57,
+      "learning_rate": 9.750000000000001e-05,
+      "loss": 2.9239,
       "step": 780
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 0.0006051484609098339,
-      "loss": 3.4596,
+      "epoch": 8.79,
+      "learning_rate": 0.0001,
+      "loss": 2.8756,
       "step": 800
     },
     {
-      "epoch": 0.43,
-      "eval_loss": Infinity,
-      "eval_runtime": 1233.785,
-      "eval_samples_per_second": 21.574,
-      "eval_steps_per_second": 0.674,
+      "epoch": 8.79,
+      "eval_loss": 2.7302000522613525,
+      "eval_runtime": 191.8065,
+      "eval_samples_per_second": 25.249,
+      "eval_steps_per_second": 0.792,
       "eval_wer": 1.0,
       "step": 800
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 0.0006010623808226641,
-      "loss": 3.2227,
+      "epoch": 9.01,
+      "learning_rate": 0.0001025,
+      "loss": 2.8933,
       "step": 820
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 0.0005969763007354945,
-      "loss": 3.2357,
+      "epoch": 9.23,
+      "learning_rate": 0.000105,
+      "loss": 2.7318,
       "step": 840
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 0.0005930945246526832,
-      "loss": 3.5563,
+      "epoch": 9.45,
+      "learning_rate": 0.0001075,
+      "loss": 2.5941,
       "step": 860
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 0.0005890084445655135,
-      "loss": 3.0845,
+      "epoch": 9.67,
+      "learning_rate": 0.00011,
+      "loss": 2.4441,
       "step": 880
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 0.0005851266684827023,
-      "loss": 3.4644,
+      "epoch": 9.89,
+      "learning_rate": 0.00011250000000000001,
+      "loss": 2.2931,
       "step": 900
     },
     {
-      "epoch": 0.48,
-      "eval_loss": Infinity,
-      "eval_runtime": 1235.1488,
-      "eval_samples_per_second": 21.55,
-      "eval_steps_per_second": 0.674,
-      "eval_wer": 1.0,
+      "epoch": 9.89,
+      "eval_loss": 1.5057899951934814,
+      "eval_runtime": 196.4368,
+      "eval_samples_per_second": 24.654,
+      "eval_steps_per_second": 0.774,
+      "eval_wer": 0.9775759296054499,
       "step": 900
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 0.0005810405883955325,
-      "loss": 3.2092,
+      "epoch": 10.11,
+      "learning_rate": 0.000115,
+      "loss": 2.1999,
       "step": 920
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 0.0005769545083083629,
-      "loss": 3.2157,
+      "epoch": 10.33,
+      "learning_rate": 0.0001175,
+      "loss": 2.0574,
       "step": 940
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 0.0005730727322255517,
-      "loss": 3.4679,
+      "epoch": 10.55,
+      "learning_rate": 0.00012,
+      "loss": 1.9672,
       "step": 960
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 0.0005689866521383819,
-      "loss": 3.0662,
+      "epoch": 10.77,
+      "learning_rate": 0.0001225,
+      "loss": 1.9015,
       "step": 980
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 0.0005651048760555707,
-      "loss": 3.4671,
+      "epoch": 10.98,
+      "learning_rate": 0.000125,
+      "loss": 1.8427,
       "step": 1000
     },
     {
-      "epoch": 0.53,
-      "eval_loss": Infinity,
-      "eval_runtime": 1241.2249,
-      "eval_samples_per_second": 21.445,
-      "eval_steps_per_second": 0.67,
-      "eval_wer": 1.0,
+      "epoch": 10.98,
+      "eval_loss": 0.9154536724090576,
+      "eval_runtime": 203.1168,
+      "eval_samples_per_second": 23.843,
+      "eval_steps_per_second": 0.748,
+      "eval_wer": 0.7832245245529378,
       "step": 1000
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 0.0005610187959684009,
-      "loss": 3.2428,
+      "epoch": 11.21,
+      "learning_rate": 0.0001275,
+      "loss": 1.8155,
       "step": 1020
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 0.0005569327158812314,
-      "loss": 3.2165,
+      "epoch": 11.43,
+      "learning_rate": 0.00013000000000000002,
+      "loss": 1.6767,
       "step": 1040
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 0.0005530509397984201,
-      "loss": 3.5106,
+      "epoch": 11.64,
+      "learning_rate": 0.00013250000000000002,
+      "loss": 1.5184,
       "step": 1060
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 0.0005489648597112503,
-      "loss": 3.1137,
+      "epoch": 11.86,
+      "learning_rate": 0.000135,
+      "loss": 1.4505,
       "step": 1080
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 0.0005452873876327976,
-      "loss": 3.6005,
+      "epoch": 12.09,
+      "learning_rate": 0.0001375,
+      "loss": 1.4286,
       "step": 1100
     },
     {
-      "epoch": 0.59,
-      "eval_loss": Infinity,
-      "eval_runtime": 1236.2955,
-      "eval_samples_per_second": 21.53,
-      "eval_steps_per_second": 0.673,
-      "eval_wer": 1.0,
+      "epoch": 12.09,
+      "eval_loss": 0.4074769914150238,
+      "eval_runtime": 194.7798,
+      "eval_samples_per_second": 24.864,
+      "eval_steps_per_second": 0.78,
+      "eval_wer": 0.37964802724950325,
       "step": 1100
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 0.0005412013075456279,
-      "loss": 3.2702,
+      "epoch": 12.31,
+      "learning_rate": 0.00014000000000000001,
+      "loss": 1.3478,
       "step": 1120
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 0.0005371152274584582,
-      "loss": 3.6768,
+      "epoch": 12.52,
+      "learning_rate": 0.0001425,
+      "loss": 1.3164,
       "step": 1140
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 0.000533233451375647,
-      "loss": 3.6313,
+      "epoch": 12.74,
+      "learning_rate": 0.000145,
+      "loss": 1.269,
       "step": 1160
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 0.0005291473712884773,
-      "loss": 3.2456,
+      "epoch": 12.96,
+      "learning_rate": 0.0001475,
+      "loss": 1.2292,
       "step": 1180
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 0.000525265595205666,
-      "loss": 3.9182,
+      "epoch": 13.19,
+      "learning_rate": 0.00015,
+      "loss": 1.2229,
       "step": 1200
     },
     {
-      "epoch": 0.64,
-      "eval_loss": Infinity,
-      "eval_runtime": 1229.4316,
-      "eval_samples_per_second": 21.651,
-      "eval_steps_per_second": 0.677,
-      "eval_wer": 1.0,
+      "epoch": 13.19,
+      "eval_loss": 0.28933778405189514,
+      "eval_runtime": 192.6398,
+      "eval_samples_per_second": 25.14,
+      "eval_steps_per_second": 0.789,
+      "eval_wer": 0.26520011353959694,
       "step": 1200
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 0.0005211795151184963,
-      "loss": 3.3805,
+      "epoch": 13.4,
+      "learning_rate": 0.0001525,
+      "loss": 1.182,
       "step": 1220
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 0.0005170934350313267,
-      "loss": 3.4687,
+      "epoch": 13.62,
+      "learning_rate": 0.000155,
+      "loss": 1.1768,
       "step": 1240
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 0.000513007354944157,
-      "loss": 3.4473,
+      "epoch": 13.84,
+      "learning_rate": 0.0001575,
+      "loss": 1.1534,
       "step": 1260
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 0.0005089212748569872,
-      "loss": 3.2232,
+      "epoch": 14.07,
+      "learning_rate": 0.00016,
+      "loss": 1.1832,
       "step": 1280
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 0.000505039498774176,
-      "loss": 3.6466,
+      "epoch": 14.28,
+      "learning_rate": 0.00016250000000000002,
+      "loss": 1.1106,
       "step": 1300
     },
     {
-      "epoch": 0.69,
-      "eval_loss": Infinity,
-      "eval_runtime": 1237.3739,
-      "eval_samples_per_second": 21.512,
-      "eval_steps_per_second": 0.672,
-      "eval_wer": 1.0,
+      "epoch": 14.28,
+      "eval_loss": 0.24685777723789215,
+      "eval_runtime": 198.5067,
+      "eval_samples_per_second": 24.397,
+      "eval_steps_per_second": 0.766,
+      "eval_wer": 0.2253760999148453,
       "step": 1300
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 0.0005009534186870062,
-      "loss": 3.3496,
+      "epoch": 14.5,
+      "learning_rate": 0.000165,
+      "loss": 1.1217,
       "step": 1320
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 0.0004968673385998365,
-      "loss": 3.3918,
+      "epoch": 14.72,
+      "learning_rate": 0.0001675,
+      "loss": 1.1236,
       "step": 1340
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 0.0004927812585126669,
-      "loss": 3.3232,
+      "epoch": 14.94,
+      "learning_rate": 0.00017,
+      "loss": 1.0969,
       "step": 1360
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 0.0004886951784254972,
-      "loss": 3.288,
+      "epoch": 15.16,
+      "learning_rate": 0.0001725,
+      "loss": 1.0971,
       "step": 1380
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 0.00048481340234268587,
-      "loss": 3.6932,
+      "epoch": 15.38,
+      "learning_rate": 0.000175,
+      "loss": 1.0663,
       "step": 1400
     },
     {
-      "epoch": 0.75,
-      "eval_loss": Infinity,
-      "eval_runtime": 1236.694,
-      "eval_samples_per_second": 21.524,
-      "eval_steps_per_second": 0.673,
-      "eval_wer": 1.0,
+      "epoch": 15.38,
+      "eval_loss": 0.22192780673503876,
+      "eval_runtime": 189.4566,
+      "eval_samples_per_second": 25.563,
+      "eval_steps_per_second": 0.802,
+      "eval_wer": 0.19727504967357365,
       "step": 1400
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 0.00048072732225551624,
-      "loss": 3.5411,
+      "epoch": 15.6,
+      "learning_rate": 0.0001775,
+      "loss": 1.0733,
       "step": 1420
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 0.0004766412421683465,
-      "loss": 3.5658,
+      "epoch": 15.82,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 1.0635,
       "step": 1440
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 0.00047275946608553524,
-      "loss": 3.7666,
+      "epoch": 16.04,
+      "learning_rate": 0.0001825,
+      "loss": 1.1079,
       "step": 1460
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 0.0004686733859983656,
-      "loss": 3.7564,
+      "epoch": 16.26,
+      "learning_rate": 0.000185,
+      "loss": 1.0467,
       "step": 1480
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 0.00046479160991555436,
-      "loss": 3.7939,
+      "epoch": 16.48,
+      "learning_rate": 0.0001875,
+      "loss": 1.0667,
       "step": 1500
     },
     {
-      "epoch": 0.8,
-      "eval_loss": Infinity,
-      "eval_runtime": 1235.7295,
-      "eval_samples_per_second": 21.54,
-      "eval_steps_per_second": 0.673,
-      "eval_wer": 1.0,
+      "epoch": 16.48,
+      "eval_loss": 0.21288961172103882,
+      "eval_runtime": 189.0078,
+      "eval_samples_per_second": 25.623,
+      "eval_steps_per_second": 0.804,
+      "eval_wer": 0.18944081748509792,
       "step": 1500
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 0.00046070552982838467,
-      "loss": 3.7471,
+      "epoch": 16.7,
+      "learning_rate": 0.00019,
+      "loss": 1.0204,
       "step": 1520
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 0.00045661944974121493,
-      "loss": 3.7457,
+      "epoch": 16.92,
+      "learning_rate": 0.00019250000000000002,
+      "loss": 1.0356,
       "step": 1540
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 0.0004527376736584037,
-      "loss": 3.798,
+      "epoch": 17.14,
+      "learning_rate": 0.00019500000000000002,
+      "loss": 1.0538,
       "step": 1560
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 0.00044865159357123405,
-      "loss": 3.7611,
+      "epoch": 17.36,
+      "learning_rate": 0.0001975,
+      "loss": 1.0228,
       "step": 1580
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 0.0004447698174884228,
-      "loss": 3.9284,
+      "epoch": 17.58,
+      "learning_rate": 0.0002,
+      "loss": 1.0193,
       "step": 1600
     },
     {
-      "epoch": 0.85,
-      "eval_loss": Infinity,
-      "eval_runtime": 1251.8158,
-      "eval_samples_per_second": 21.264,
-      "eval_steps_per_second": 0.665,
-      "eval_wer": 1.0,
+      "epoch": 17.58,
+      "eval_loss": 0.1991206556558609,
+      "eval_runtime": 191.1717,
+      "eval_samples_per_second": 25.333,
+      "eval_steps_per_second": 0.795,
+      "eval_wer": 0.17885325007096226,
       "step": 1600
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 0.0004406837374012531,
-      "loss": 3.7416,
+      "epoch": 17.8,
+      "learning_rate": 0.00020250000000000002,
+      "loss": 1.0322,
       "step": 1620
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 0.00043659765731408337,
-      "loss": 3.7416,
+      "epoch": 18.02,
+      "learning_rate": 0.000205,
+      "loss": 1.0176,
       "step": 1640
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 0.0004327158812312721,
-      "loss": 3.8024,
+      "epoch": 18.24,
+      "learning_rate": 0.0002075,
+      "loss": 1.0272,
       "step": 1660
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 0.0004286298011441025,
-      "loss": 3.7578,
+      "epoch": 18.46,
+      "learning_rate": 0.00021,
+      "loss": 0.9675,
       "step": 1680
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 0.0004247480250612912,
-      "loss": 3.7859,
+      "epoch": 18.68,
+      "learning_rate": 0.0002125,
+      "loss": 0.9816,
       "step": 1700
     },
     {
-      "epoch": 0.91,
-      "eval_loss": Infinity,
-      "eval_runtime": 1235.734,
-      "eval_samples_per_second": 21.54,
-      "eval_steps_per_second": 0.673,
-      "eval_wer": 1.0,
+      "epoch": 18.68,
+      "eval_loss": 0.19402356445789337,
+      "eval_runtime": 194.75,
+      "eval_samples_per_second": 24.868,
+      "eval_steps_per_second": 0.78,
+      "eval_wer": 0.18010218563724098,
       "step": 1700
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 0.0004206619449741215,
-      "loss": 3.7427,
+      "epoch": 18.9,
+      "learning_rate": 0.000215,
+      "loss": 1.0023,
       "step": 1720
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 0.0004165758648869518,
-      "loss": 3.751,
+      "epoch": 19.12,
+      "learning_rate": 0.0002175,
+      "loss": 1.013,
       "step": 1740
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 0.00041248978479978206,
-      "loss": 3.6701,
+      "epoch": 19.34,
+      "learning_rate": 0.00022,
+      "loss": 0.9664,
       "step": 1760
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 0.0004084037047126123,
-      "loss": 3.7613,
+      "epoch": 19.56,
+      "learning_rate": 0.00022250000000000001,
+      "loss": 0.9736,
       "step": 1780
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 0.00040472623263415966,
-      "loss": 3.9363,
+      "epoch": 19.78,
+      "learning_rate": 0.00022500000000000002,
+      "loss": 0.9814,
       "step": 1800
     },
     {
-      "epoch": 0.96,
-      "eval_loss": Infinity,
-      "eval_runtime": 1236.9158,
-      "eval_samples_per_second": 21.52,
-      "eval_steps_per_second": 0.673,
-      "eval_wer": 1.0,
+      "epoch": 19.78,
+      "eval_loss": 0.18596723675727844,
+      "eval_runtime": 188.4639,
+      "eval_samples_per_second": 25.697,
+      "eval_steps_per_second": 0.807,
+      "eval_wer": 0.16667612829974454,
       "step": 1800
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 0.0004006401525469899,
-      "loss": 3.7464,
+      "epoch": 19.99,
+      "learning_rate": 0.0002275,
+      "loss": 1.0064,
       "step": 1820
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 0.00039655407245982023,
-      "loss": 3.7414,
+      "epoch": 20.22,
+      "learning_rate": 0.00023,
+      "loss": 0.9583,
       "step": 1840
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 0.000392672296377009,
-      "loss": 3.8036,
+      "epoch": 20.44,
+      "learning_rate": 0.0002325,
+      "loss": 0.9646,
       "step": 1860
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 0.00038879052029419783,
-      "loss": 3.8768,
+      "epoch": 20.66,
+      "learning_rate": 0.000235,
+      "loss": 0.9762,
       "step": 1880
     },
     {
-      "epoch": 1.01,
-      "learning_rate": 0.0003847044402070281,
-      "loss": 3.7573,
+      "epoch": 20.87,
+      "learning_rate": 0.0002375,
+      "loss": 0.9787,
       "step": 1900
     },
     {
-      "epoch": 1.01,
-      "eval_loss": Infinity,
-      "eval_runtime": 1232.3535,
-      "eval_samples_per_second": 21.599,
-      "eval_steps_per_second": 0.675,
-      "eval_wer": 1.0,
+      "epoch": 20.87,
+      "eval_loss": 0.18878202140331268,
+      "eval_runtime": 188.244,
+      "eval_samples_per_second": 25.727,
+      "eval_steps_per_second": 0.807,
+      "eval_wer": 0.16420664206642066,
       "step": 1900
     },
     {
-      "epoch": 1.03,
-      "learning_rate": 0.00038061836011985835,
-      "loss": 3.6959,
+      "epoch": 21.1,
+      "learning_rate": 0.00024,
+      "loss": 1.0218,
       "step": 1920
     },
     {
-      "epoch": 1.04,
-      "learning_rate": 0.00037653228003268867,
-      "loss": 3.7313,
+      "epoch": 21.32,
+      "learning_rate": 0.00024249999999999999,
+      "loss": 0.9505,
       "step": 1940
     },
     {
-      "epoch": 1.05,
-      "learning_rate": 0.00037244619994551893,
-      "loss": 3.7676,
+      "epoch": 21.54,
+      "learning_rate": 0.000245,
+      "loss": 0.9554,
       "step": 1960
     },
     {
-      "epoch": 1.06,
-      "learning_rate": 0.0003685644238627077,
-      "loss": 3.7872,
+      "epoch": 21.75,
+      "learning_rate": 0.0002475,
+      "loss": 0.9728,
       "step": 1980
     },
     {
-      "epoch": 1.07,
-      "learning_rate": 0.000364478343775538,
-      "loss": 3.7553,
+      "epoch": 21.97,
+      "learning_rate": 0.00025,
+      "loss": 0.9699,
       "step": 2000
     },
     {
-      "epoch": 1.07,
-      "eval_loss": Infinity,
-      "eval_runtime": 1239.2247,
-      "eval_samples_per_second": 21.48,
-      "eval_steps_per_second": 0.671,
-      "eval_wer": 1.0,
+      "epoch": 21.97,
+      "eval_loss": 0.18748582899570465,
+      "eval_runtime": 190.7875,
+      "eval_samples_per_second": 25.384,
+      "eval_steps_per_second": 0.797,
+      "eval_wer": 0.17042293499858074,
       "step": 2000
     },
     {
-      "epoch": 1.08,
-      "learning_rate": 0.0003603922636883683,
-      "loss": 3.694,
+      "epoch": 22.2,
+      "learning_rate": 0.00024803921568627453,
+      "loss": 0.9624,
       "step": 2020
     },
     {
-      "epoch": 1.09,
-      "learning_rate": 0.0003565104876055571,
-      "loss": 3.8389,
+      "epoch": 22.42,
+      "learning_rate": 0.000246078431372549,
+      "loss": 0.9419,
       "step": 2040
     },
     {
-      "epoch": 1.1,
-      "learning_rate": 0.00035242440751838736,
-      "loss": 3.7573,
+      "epoch": 22.63,
+      "learning_rate": 0.00024411764705882354,
+      "loss": 0.9563,
       "step": 2060
     },
     {
-      "epoch": 1.11,
-      "learning_rate": 0.0003483383274312177,
-      "loss": 3.6641,
+      "epoch": 22.85,
+      "learning_rate": 0.00024215686274509804,
+      "loss": 0.9643,
       "step": 2080
     },
     {
-      "epoch": 1.12,
-      "learning_rate": 0.00034425224734404794,
-      "loss": 3.7606,
+      "epoch": 23.08,
+      "learning_rate": 0.00024019607843137256,
+      "loss": 0.9616,
       "step": 2100
     },
     {
-      "epoch": 1.12,
-      "eval_loss": Infinity,
-      "eval_runtime": 1237.7997,
-      "eval_samples_per_second": 21.504,
-      "eval_steps_per_second": 0.672,
-      "eval_wer": 1.0,
+      "epoch": 23.08,
+      "eval_loss": 0.18017500638961792,
+      "eval_runtime": 191.3932,
+      "eval_samples_per_second": 25.304,
+      "eval_steps_per_second": 0.794,
+      "eval_wer": 0.16173715583309678,
       "step": 2100
     },
     {
-      "epoch": 1.13,
-      "learning_rate": 0.00034016616725687825,
-      "loss": 3.6827,
+      "epoch": 23.3,
+      "learning_rate": 0.00023823529411764704,
+      "loss": 0.917,
       "step": 2120
     },
     {
-      "epoch": 1.14,
-      "learning_rate": 0.000336284391174067,
-      "loss": 3.8375,
+      "epoch": 23.51,
+      "learning_rate": 0.00023627450980392157,
+      "loss": 0.945,
       "step": 2140
     },
     {
-      "epoch": 1.15,
-      "learning_rate": 0.0003321983110868973,
-      "loss": 3.7527,
+      "epoch": 23.73,
+      "learning_rate": 0.0002343137254901961,
+      "loss": 0.9243,
       "step": 2160
     },
     {
-      "epoch": 1.16,
-      "learning_rate": 0.0003283165350040861,
-      "loss": 3.7881,
+      "epoch": 23.95,
+      "learning_rate": 0.0002323529411764706,
+      "loss": 0.9288,
       "step": 2180
     },
     {
-      "epoch": 1.17,
-      "learning_rate": 0.0003242304549169164,
-      "loss": 3.7514,
+      "epoch": 24.17,
+      "learning_rate": 0.0002303921568627451,
+      "loss": 0.9378,
       "step": 2200
     },
     {
-      "epoch": 1.17,
-      "eval_loss": Infinity,
-      "eval_runtime": 1237.076,
-      "eval_samples_per_second": 21.517,
-      "eval_steps_per_second": 0.673,
-      "eval_wer": 1.0,
+      "epoch": 24.17,
+      "eval_loss": 0.17928896844387054,
+      "eval_runtime": 189.7619,
+      "eval_samples_per_second": 25.521,
+      "eval_steps_per_second": 0.801,
+      "eval_wer": 0.1577348850411581,
       "step": 2200
     },
     {
-      "epoch": 1.19,
-      "learning_rate": 0.0003201443748297467,
-      "loss": 3.6727,
+      "epoch": 24.39,
+      "learning_rate": 0.0002284313725490196,
+      "loss": 0.9071,
       "step": 2220
     },
     {
-      "epoch": 1.2,
-      "learning_rate": 0.00031626259874693543,
-      "loss": 3.8424,
+      "epoch": 24.61,
+      "learning_rate": 0.00022647058823529412,
+      "loss": 0.9054,
       "step": 2240
     },
     {
-      "epoch": 1.21,
-      "learning_rate": 0.00031217651865976574,
-      "loss": 3.7594,
+      "epoch": 24.83,
+      "learning_rate": 0.0002246078431372549,
+      "loss": 0.9303,
       "step": 2260
     },
     {
-      "epoch": 1.22,
-      "learning_rate": 0.00030829474257695454,
-      "loss": 3.7862,
+      "epoch": 25.05,
+      "learning_rate": 0.00022264705882352943,
+      "loss": 0.9376,
       "step": 2280
     },
     {
-      "epoch": 1.23,
-      "learning_rate": 0.0003042086624897848,
-      "loss": 3.7472,
+      "epoch": 25.27,
+      "learning_rate": 0.0002206862745098039,
+      "loss": 0.888,
       "step": 2300
     },
     {
-      "epoch": 1.23,
-      "eval_loss": Infinity,
-      "eval_runtime": 1236.6319,
-      "eval_samples_per_second": 21.525,
-      "eval_steps_per_second": 0.673,
-      "eval_wer": 1.0,
+      "epoch": 25.27,
+      "eval_loss": 0.17642559111118317,
+      "eval_runtime": 187.0437,
+      "eval_samples_per_second": 25.892,
+      "eval_steps_per_second": 0.813,
+      "eval_wer": 0.15452739142776042,
       "step": 2300
     },
     {
-      "epoch": 1.24,
-      "learning_rate": 0.0003001225824026151,
-      "loss": 3.6956,
+      "epoch": 25.49,
+      "learning_rate": 0.00021872549019607843,
+      "loss": 0.9135,
       "step": 2320
     },
     {
-      "epoch": 1.25,
-      "learning_rate": 0.0002960365023154454,
-      "loss": 3.7248,
+      "epoch": 25.71,
+      "learning_rate": 0.00021676470588235294,
+      "loss": 0.9094,
       "step": 2340
     },
     {
-      "epoch": 1.26,
-      "learning_rate": 0.0002919504222282757,
-      "loss": 3.7531,
+      "epoch": 25.93,
+      "learning_rate": 0.00021480392156862746,
+      "loss": 0.8879,
       "step": 2360
     },
     {
-      "epoch": 1.27,
-      "learning_rate": 0.0002882729501498229,
-      "loss": 3.9245,
+      "epoch": 26.15,
+      "learning_rate": 0.00021284313725490196,
+      "loss": 0.929,
       "step": 2380
     },
     {
-      "epoch": 1.28,
-      "learning_rate": 0.00028418687006265324,
-      "loss": 3.7478,
+      "epoch": 26.37,
+      "learning_rate": 0.00021088235294117647,
+      "loss": 0.8942,
       "step": 2400
     },
     {
-      "epoch": 1.28,
-      "eval_loss": Infinity,
-      "eval_runtime": 1236.7725,
-      "eval_samples_per_second": 21.522,
-      "eval_steps_per_second": 0.673,
-      "eval_wer": 1.0,
+      "epoch": 26.37,
+      "eval_loss": 0.16744859516620636,
+      "eval_runtime": 190.6796,
+      "eval_samples_per_second": 25.399,
+      "eval_steps_per_second": 0.797,
+      "eval_wer": 0.14916264547260857,
       "step": 2400
     },
     {
-      "epoch": 1.29,
-      "learning_rate": 0.00028010078997548355,
-      "loss": 3.6826,
+      "epoch": 26.59,
+      "learning_rate": 0.000208921568627451,
+      "loss": 0.8717,
       "step": 2420
     },
     {
-      "epoch": 1.3,
-      "learning_rate": 0.0002760147098883138,
-      "loss": 3.7236,
+      "epoch": 26.81,
+      "learning_rate": 0.0002069607843137255,
+      "loss": 0.8952,
       "step": 2440
     },
     {
-      "epoch": 1.31,
-      "learning_rate": 0.00027192862980114413,
-      "loss": 3.7609,
+      "epoch": 27.03,
+      "learning_rate": 0.000205,
+      "loss": 0.8849,
       "step": 2460
     },
     {
-      "epoch": 1.32,
-      "learning_rate": 0.00026804685371833287,
-      "loss": 3.7846,
+      "epoch": 27.25,
+      "learning_rate": 0.00020313725490196078,
+      "loss": 0.8781,
       "step": 2480
     },
     {
-      "epoch": 1.33,
-      "learning_rate": 0.0002639607736311632,
-      "loss": 3.7496,
+      "epoch": 27.47,
+      "learning_rate": 0.0002011764705882353,
+      "loss": 0.8701,
       "step": 2500
     },
     {
-      "epoch": 1.33,
-      "eval_loss": Infinity,
-      "eval_runtime": 1233.5496,
-      "eval_samples_per_second": 21.578,
-      "eval_steps_per_second": 0.674,
-      "eval_wer": 1.0,
+      "epoch": 27.47,
+      "eval_loss": 0.1738910973072052,
+      "eval_runtime": 187.7205,
+      "eval_samples_per_second": 25.799,
+      "eval_steps_per_second": 0.81,
+      "eval_wer": 0.1511779733181947,
       "step": 2500
     },
     {
-      "epoch": 1.35,
-      "learning_rate": 0.00025987469354399345,
-      "loss": 3.6785,
+      "epoch": 27.69,
+      "learning_rate": 0.0001992156862745098,
+      "loss": 0.8674,
       "step": 2520
     },
     {
-      "epoch": 1.36,
-      "learning_rate": 0.00025578861345682376,
-      "loss": 3.7212,
+      "epoch": 27.91,
+      "learning_rate": 0.00019725490196078433,
+      "loss": 0.8752,
       "step": 2540
     },
     {
-      "epoch": 1.37,
-      "learning_rate": 0.0002517025333696541,
-      "loss": 3.7637,
+      "epoch": 28.13,
+      "learning_rate": 0.0001952941176470588,
+      "loss": 0.91,
       "step": 2560
     },
     {
-      "epoch": 1.38,
-      "learning_rate": 0.00024761645328248434,
-      "loss": 3.6513,
+      "epoch": 28.35,
+      "learning_rate": 0.00019333333333333333,
+      "loss": 0.8693,
       "step": 2580
     },
     {
-      "epoch": 1.39,
-      "learning_rate": 0.00024353037319531465,
-      "loss": 3.7513,
+      "epoch": 28.57,
+      "learning_rate": 0.00019137254901960786,
+      "loss": 0.8555,
       "step": 2600
     },
     {
-      "epoch": 1.39,
-      "eval_loss": Infinity,
-      "eval_runtime": 1231.8132,
-      "eval_samples_per_second": 21.609,
-      "eval_steps_per_second": 0.675,
-      "eval_wer": 1.0,
+      "epoch": 28.57,
+      "eval_loss": 0.1689654141664505,
+      "eval_runtime": 196.2665,
+      "eval_samples_per_second": 24.676,
+      "eval_steps_per_second": 0.774,
+      "eval_wer": 0.14459267669599773,
       "step": 2600
     },
     {
-      "epoch": 1.4,
-      "learning_rate": 0.00023944429310814494,
-      "loss": 3.6818,
+      "epoch": 28.79,
+      "learning_rate": 0.00018941176470588236,
+      "loss": 0.8796,
       "step": 2620
     },
     {
-      "epoch": 1.41,
-      "learning_rate": 0.00023556251702533368,
-      "loss": 3.8417,
+      "epoch": 29.01,
+      "learning_rate": 0.00018745098039215686,
+      "loss": 0.9045,
       "step": 2640
     },
     {
-      "epoch": 1.42,
-      "learning_rate": 0.000231476436938164,
-      "loss": 3.7639,
+      "epoch": 29.23,
+      "learning_rate": 0.00018549019607843137,
+      "loss": 0.8515,
       "step": 2660
     },
     {
-      "epoch": 1.43,
-      "learning_rate": 0.00022759466085535277,
-      "loss": 3.7842,
+      "epoch": 29.45,
+      "learning_rate": 0.0001835294117647059,
+      "loss": 0.861,
       "step": 2680
     },
     {
-      "epoch": 1.44,
-      "learning_rate": 0.00022350858076818309,
-      "loss": 3.7497,
+      "epoch": 29.67,
+      "learning_rate": 0.0001815686274509804,
+      "loss": 0.8513,
       "step": 2700
     },
     {
-      "epoch": 1.44,
-      "eval_loss": Infinity,
-      "eval_runtime": 1234.8437,
-      "eval_samples_per_second": 21.556,
-      "eval_steps_per_second": 0.674,
-      "eval_wer": 1.0,
+      "epoch": 29.67,
+      "eval_loss": 0.16488835215568542,
+      "eval_runtime": 189.1938,
+      "eval_samples_per_second": 25.598,
+      "eval_steps_per_second": 0.803,
+      "eval_wer": 0.14774340051092819,
       "step": 2700
     },
     {
-      "epoch": 1.45,
-      "learning_rate": 0.00021942250068101335,
-      "loss": 3.6795,
+      "epoch": 29.89,
+      "learning_rate": 0.0001796078431372549,
+      "loss": 0.8573,
       "step": 2720
     },
     {
-      "epoch": 1.46,
-      "learning_rate": 0.00021554072459820212,
-      "loss": 3.8421,
+      "epoch": 30.11,
+      "learning_rate": 0.00017764705882352942,
+      "loss": 0.8686,
       "step": 2740
     },
     {
-      "epoch": 1.47,
-      "learning_rate": 0.00021145464451103243,
-      "loss": 3.7614,
+      "epoch": 30.33,
+      "learning_rate": 0.00017568627450980392,
+      "loss": 0.8462,
       "step": 2760
     },
     {
-      "epoch": 1.48,
-      "learning_rate": 0.0002075728684282212,
-      "loss": 3.9259,
+      "epoch": 30.55,
+      "learning_rate": 0.00017372549019607845,
+      "loss": 0.8419,
       "step": 2780
     },
     {
-      "epoch": 1.49,
-      "learning_rate": 0.00020348678834105147,
-      "loss": 3.7539,
+      "epoch": 30.77,
+      "learning_rate": 0.00017176470588235293,
+      "loss": 0.8659,
       "step": 2800
     },
     {
-      "epoch": 1.49,
-      "eval_loss": Infinity,
-      "eval_runtime": 1240.1103,
-      "eval_samples_per_second": 21.464,
-      "eval_steps_per_second": 0.671,
-      "eval_wer": 1.0,
+      "epoch": 30.77,
+      "eval_loss": 0.1636602133512497,
+      "eval_runtime": 190.245,
+      "eval_samples_per_second": 25.457,
+      "eval_steps_per_second": 0.799,
+      "eval_wer": 0.1422367300596083,
       "step": 2800
     },
     {
-      "epoch": 1.51,
-      "learning_rate": 0.00019940070825388178,
-      "loss": 3.6889,
+      "epoch": 30.98,
+      "learning_rate": 0.00016980392156862745,
+      "loss": 0.8607,
       "step": 2820
     },
     {
-      "epoch": 1.52,
-      "learning_rate": 0.00019551893217107055,
-      "loss": 3.843,
+      "epoch": 31.21,
+      "learning_rate": 0.00016784313725490196,
+      "loss": 0.8484,
       "step": 2840
     },
     {
-      "epoch": 1.53,
-      "learning_rate": 0.00019143285208390087,
-      "loss": 3.7666,
+      "epoch": 31.43,
+      "learning_rate": 0.00016588235294117648,
+      "loss": 0.8335,
       "step": 2860
     },
     {
-      "epoch": 1.54,
-      "learning_rate": 0.00018755107600108964,
-      "loss": 3.786,
+      "epoch": 31.64,
+      "learning_rate": 0.00016392156862745098,
+      "loss": 0.8321,
       "step": 2880
     },
     {
-      "epoch": 1.55,
-      "learning_rate": 0.0001834649959139199,
-      "loss": 3.7581,
+      "epoch": 31.86,
+      "learning_rate": 0.00016196078431372549,
+      "loss": 0.8419,
       "step": 2900
     },
     {
-      "epoch": 1.55,
-      "eval_loss": Infinity,
-      "eval_runtime": 1231.1699,
-      "eval_samples_per_second": 21.62,
-      "eval_steps_per_second": 0.676,
-      "eval_wer": 1.0,
+      "epoch": 31.86,
+      "eval_loss": 0.16137854754924774,
+      "eval_runtime": 190.2053,
+      "eval_samples_per_second": 25.462,
+      "eval_steps_per_second": 0.799,
+      "eval_wer": 0.1397104740278172,
       "step": 2900
     },
     {
-      "epoch": 1.56,
-      "learning_rate": 0.00017937891582675021,
-      "loss": 3.672,
+      "epoch": 32.09,
+      "learning_rate": 0.00016,
+      "loss": 0.8634,
       "step": 2920
     },
     {
-      "epoch": 1.57,
-      "learning_rate": 0.00017549713974393899,
-      "loss": 3.8539,
+      "epoch": 32.31,
+      "learning_rate": 0.00015803921568627451,
+      "loss": 0.8351,
       "step": 2940
     },
     {
-      "epoch": 1.58,
-      "learning_rate": 0.00017141105965676927,
-      "loss": 3.7652,
+      "epoch": 32.52,
+      "learning_rate": 0.00015607843137254901,
+      "loss": 0.8427,
       "step": 2960
     },
     {
-      "epoch": 1.59,
-      "learning_rate": 0.00016752928357395807,
-      "loss": 3.7825,
+      "epoch": 32.74,
+      "learning_rate": 0.00015411764705882352,
+      "loss": 0.8324,
       "step": 2980
     },
     {
-      "epoch": 1.6,
-      "learning_rate": 0.00016344320348678833,
-      "loss": 3.7572,
+      "epoch": 32.96,
+      "learning_rate": 0.00015215686274509804,
+      "loss": 0.8491,
       "step": 3000
     },
     {
-      "epoch": 1.6,
-      "eval_loss": Infinity,
-      "eval_runtime": 1241.8499,
-      "eval_samples_per_second": 21.434,
-      "eval_steps_per_second": 0.67,
-      "eval_wer": 1.0,
+      "epoch": 32.96,
+      "eval_loss": 0.1594749242067337,
+      "eval_runtime": 191.4631,
+      "eval_samples_per_second": 25.295,
+      "eval_steps_per_second": 0.794,
+      "eval_wer": 0.1401362475163213,
       "step": 3000
     },
     {
-      "epoch": 1.61,
-      "learning_rate": 0.00015935712339961862,
-      "loss": 3.6906,
+      "epoch": 33.19,
+      "learning_rate": 0.00015019607843137257,
+      "loss": 0.8263,
       "step": 3020
     },
     {
-      "epoch": 1.62,
-      "learning_rate": 0.00015547534731680742,
-      "loss": 3.845,
+      "epoch": 33.4,
+      "learning_rate": 0.00014823529411764705,
+      "loss": 0.8203,
       "step": 3040
     },
     {
-      "epoch": 1.63,
-      "learning_rate": 0.0001513892672296377,
-      "loss": 3.763,
+      "epoch": 33.62,
+      "learning_rate": 0.00014627450980392157,
+      "loss": 0.8277,
       "step": 3060
     },
     {
-      "epoch": 1.64,
-      "learning_rate": 0.00014750749114682648,
-      "loss": 3.779,
+      "epoch": 33.84,
+      "learning_rate": 0.00014431372549019607,
+      "loss": 0.8183,
       "step": 3080
     },
     {
-      "epoch": 1.66,
-      "learning_rate": 0.0001434214110596568,
-      "loss": 3.7589,
+      "epoch": 34.07,
+      "learning_rate": 0.0001423529411764706,
+      "loss": 0.8395,
       "step": 3100
     },
     {
-      "epoch": 1.66,
-      "eval_loss": Infinity,
-      "eval_runtime": 1236.916,
-      "eval_samples_per_second": 21.52,
-      "eval_steps_per_second": 0.673,
-      "eval_wer": 1.0,
+      "epoch": 34.07,
+      "eval_loss": 0.1606692373752594,
+      "eval_runtime": 190.8222,
+      "eval_samples_per_second": 25.38,
+      "eval_steps_per_second": 0.797,
+      "eval_wer": 0.13755322168606302,
       "step": 3100
     },
     {
-      "epoch": 1.67,
-      "learning_rate": 0.00013933533097248705,
-      "loss": 3.6948,
+      "epoch": 34.28,
+      "learning_rate": 0.0001403921568627451,
+      "loss": 0.8207,
       "step": 3120
     },
     {
-      "epoch": 1.68,
-      "learning_rate": 0.00013524925088531734,
-      "loss": 3.7211,
+      "epoch": 34.5,
+      "learning_rate": 0.0001384313725490196,
+      "loss": 0.8106,
       "step": 3140
     },
     {
-      "epoch": 1.69,
-      "learning_rate": 0.00013116317079814763,
-      "loss": 3.7588,
+      "epoch": 34.72,
+      "learning_rate": 0.00013647058823529413,
+      "loss": 0.8329,
       "step": 3160
     },
     {
-      "epoch": 1.7,
-      "learning_rate": 0.00012728139471533643,
-      "loss": 3.7898,
+      "epoch": 34.94,
+      "learning_rate": 0.00013450980392156863,
+      "loss": 0.8048,
       "step": 3180
     },
     {
-      "epoch": 1.71,
-      "learning_rate": 0.00012319531462816672,
-      "loss": 3.7592,
+      "epoch": 35.16,
+      "learning_rate": 0.00013254901960784313,
+      "loss": 0.83,
       "step": 3200
     },
     {
-      "epoch": 1.71,
-      "eval_loss": Infinity,
-      "eval_runtime": 1232.1567,
-      "eval_samples_per_second": 21.603,
-      "eval_steps_per_second": 0.675,
-      "eval_wer": 1.0,
+      "epoch": 35.16,
+      "eval_loss": 0.15384173393249512,
+      "eval_runtime": 191.1912,
+      "eval_samples_per_second": 25.331,
+      "eval_steps_per_second": 0.795,
+      "eval_wer": 0.1379222253760999,
       "step": 3200
     },
     {
-      "epoch": 1.72,
-      "learning_rate": 0.000119109234540997,
-      "loss": 3.688,
+      "epoch": 35.38,
+      "learning_rate": 0.00013058823529411764,
+      "loss": 0.7963,
       "step": 3220
     },
     {
-      "epoch": 1.73,
-      "learning_rate": 0.00011522745845818579,
-      "loss": 3.8419,
+      "epoch": 35.6,
+      "learning_rate": 0.00012862745098039216,
+      "loss": 0.7895,
       "step": 3240
     },
     {
-      "epoch": 1.74,
-      "learning_rate": 0.00011114137837101608,
-      "loss": 3.7578,
+      "epoch": 35.82,
+      "learning_rate": 0.0001266666666666667,
+      "loss": 0.7964,
       "step": 3260
     },
     {
-      "epoch": 1.75,
-      "learning_rate": 0.00010705529828384636,
-      "loss": 3.6534,
+      "epoch": 36.04,
+      "learning_rate": 0.0001247058823529412,
+      "loss": 0.7931,
       "step": 3280
     },
     {
-      "epoch": 1.76,
-      "learning_rate": 0.00010296921819667667,
-      "loss": 3.7531,
+      "epoch": 36.26,
+      "learning_rate": 0.0001227450980392157,
+      "loss": 0.7835,
       "step": 3300
     },
     {
-      "epoch": 1.76,
-      "eval_loss": Infinity,
-      "eval_runtime": 1237.4955,
-      "eval_samples_per_second": 21.51,
-      "eval_steps_per_second": 0.672,
-      "eval_wer": 1.0,
+      "epoch": 36.26,
+      "eval_loss": 0.16020993888378143,
+      "eval_runtime": 192.1405,
+      "eval_samples_per_second": 25.206,
+      "eval_steps_per_second": 0.791,
+      "eval_wer": 0.1408458699971615,
       "step": 3300
     },
     {
-      "epoch": 1.77,
-      "learning_rate": 9.888313810950695e-05,
-      "loss": 3.691,
+      "epoch": 36.48,
+      "learning_rate": 0.00012078431372549021,
+      "loss": 0.772,
       "step": 3320
     },
     {
-      "epoch": 1.78,
-      "learning_rate": 9.500136202669572e-05,
-      "loss": 3.8424,
+      "epoch": 36.7,
+      "learning_rate": 0.00011882352941176471,
+      "loss": 0.7818,
       "step": 3340
     },
     {
-      "epoch": 1.79,
-      "learning_rate": 9.091528193952601e-05,
-      "loss": 3.7698,
+      "epoch": 36.92,
+      "learning_rate": 0.00011696078431372549,
+      "loss": 0.8016,
       "step": 3360
     },
     {
-      "epoch": 1.8,
-      "learning_rate": 8.70335058567148e-05,
-      "loss": 3.7836,
+      "epoch": 37.14,
+      "learning_rate": 0.000115,
+      "loss": 0.8061,
       "step": 3380
     },
     {
-      "epoch": 1.82,
-      "learning_rate": 8.294742576954509e-05,
-      "loss": 3.7567,
+      "epoch": 37.36,
+      "learning_rate": 0.0001130392156862745,
+      "loss": 0.7703,
       "step": 3400
     },
     {
-      "epoch": 1.82,
-      "eval_loss": Infinity,
-      "eval_runtime": 1242.7888,
-      "eval_samples_per_second": 21.418,
-      "eval_steps_per_second": 0.669,
-      "eval_wer": 1.0,
+      "epoch": 37.36,
+      "eval_loss": 0.16011376678943634,
+      "eval_runtime": 187.5367,
+      "eval_samples_per_second": 25.824,
+      "eval_steps_per_second": 0.811,
+      "eval_wer": 0.13692875390292364,
       "step": 3400
     },
     {
-      "epoch": 1.83,
-      "learning_rate": 7.886134568237537e-05,
-      "loss": 3.6741,
+      "epoch": 37.58,
+      "learning_rate": 0.00011107843137254903,
+      "loss": 0.7713,
       "step": 3420
     },
     {
-      "epoch": 1.84,
-      "learning_rate": 7.497956959956416e-05,
-      "loss": 3.8442,
+      "epoch": 37.8,
+      "learning_rate": 0.00010911764705882353,
+      "loss": 0.7712,
       "step": 3440
     },
     {
-      "epoch": 1.85,
-      "learning_rate": 7.089348951239445e-05,
-      "loss": 3.7663,
+      "epoch": 38.02,
+      "learning_rate": 0.00010715686274509805,
+      "loss": 0.8149,
       "step": 3460
     },
     {
-      "epoch": 1.86,
-      "learning_rate": 6.701171342958322e-05,
-      "loss": 3.7789,
+      "epoch": 38.24,
+      "learning_rate": 0.00010519607843137255,
+      "loss": 0.7885,
       "step": 3480
     },
     {
-      "epoch": 1.87,
-      "learning_rate": 6.292563334241352e-05,
-      "loss": 3.7613,
+      "epoch": 38.46,
+      "learning_rate": 0.00010323529411764706,
+      "loss": 0.7474,
       "step": 3500
     },
     {
-      "epoch": 1.87,
-      "eval_loss": Infinity,
-      "eval_runtime": 1244.1733,
-      "eval_samples_per_second": 21.394,
-      "eval_steps_per_second": 0.669,
-      "eval_wer": 1.0,
+      "epoch": 38.46,
+      "eval_loss": 0.1514146625995636,
+      "eval_runtime": 190.8228,
+      "eval_samples_per_second": 25.38,
+      "eval_steps_per_second": 0.797,
+      "eval_wer": 0.1342321884757309,
       "step": 3500
     },
     {
-      "epoch": 1.88,
-      "learning_rate": 5.88395532552438e-05,
-      "loss": 3.6884,
+      "epoch": 38.68,
+      "learning_rate": 0.00010127450980392156,
+      "loss": 0.785,
       "step": 3520
     },
     {
-      "epoch": 1.89,
-      "learning_rate": 5.4957777172432585e-05,
-      "loss": 3.8268,
+      "epoch": 38.9,
+      "learning_rate": 9.931372549019609e-05,
+      "loss": 0.7677,
       "step": 3540
     },
     {
-      "epoch": 1.9,
-      "learning_rate": 5.0871697085262866e-05,
-      "loss": 3.7517,
+      "epoch": 39.12,
+      "learning_rate": 9.73529411764706e-05,
+      "loss": 0.7849,
       "step": 3560
     },
     {
-      "epoch": 1.91,
-      "learning_rate": 4.6989921002451645e-05,
-      "loss": 3.7802,
+      "epoch": 39.34,
+      "learning_rate": 9.539215686274511e-05,
+      "loss": 0.7637,
       "step": 3580
     },
     {
-      "epoch": 1.92,
-      "learning_rate": 4.290384091528194e-05,
-      "loss": 3.7516,
+      "epoch": 39.56,
+      "learning_rate": 9.343137254901961e-05,
+      "loss": 0.7719,
       "step": 3600
     },
     {
-      "epoch": 1.92,
-      "eval_loss": Infinity,
-      "eval_runtime": 1253.9236,
-      "eval_samples_per_second": 21.228,
-      "eval_steps_per_second": 0.664,
-      "eval_wer": 1.0,
+      "epoch": 39.56,
+      "eval_loss": 0.15932896733283997,
+      "eval_runtime": 189.6806,
+      "eval_samples_per_second": 25.532,
+      "eval_steps_per_second": 0.801,
+      "eval_wer": 0.1352540448481408,
       "step": 3600
     },
     {
-      "epoch": 1.93,
-      "learning_rate": 3.8817760828112234e-05,
-      "loss": 3.6719,
+      "epoch": 39.78,
+      "learning_rate": 9.147058823529412e-05,
+      "loss": 0.7591,
       "step": 3620
     },
     {
-      "epoch": 1.94,
-      "learning_rate": 3.4935984745301006e-05,
-      "loss": 3.8471,
+      "epoch": 39.99,
+      "learning_rate": 8.950980392156862e-05,
+      "loss": 0.7706,
       "step": 3640
     },
     {
-      "epoch": 1.95,
-      "learning_rate": 3.08499046581313e-05,
-      "loss": 3.7568,
+      "epoch": 40.22,
+      "learning_rate": 8.754901960784314e-05,
+      "loss": 0.7805,
       "step": 3660
     },
     {
-      "epoch": 1.96,
-      "learning_rate": 2.6968128575320075e-05,
-      "loss": 3.7887,
+      "epoch": 40.44,
+      "learning_rate": 8.558823529411765e-05,
+      "loss": 0.7753,
       "step": 3680
     },
     {
-      "epoch": 1.98,
-      "learning_rate": 2.2882048488150366e-05,
-      "loss": 3.7581,
+      "epoch": 40.66,
+      "learning_rate": 8.362745098039217e-05,
+      "loss": 0.7638,
       "step": 3700
     },
     {
-      "epoch": 1.98,
-      "eval_loss": Infinity,
-      "eval_runtime": 1242.0033,
-      "eval_samples_per_second": 21.432,
-      "eval_steps_per_second": 0.67,
-      "eval_wer": 1.0,
+      "epoch": 40.66,
+      "eval_loss": 0.15362653136253357,
+      "eval_runtime": 192.1118,
+      "eval_samples_per_second": 25.209,
+      "eval_steps_per_second": 0.791,
+      "eval_wer": 0.13380641498722678,
       "step": 3700
     },
     {
-      "epoch": 1.99,
-      "learning_rate": 1.879596840098066e-05,
-      "loss": 3.6785,
+      "epoch": 40.87,
+      "learning_rate": 8.166666666666667e-05,
+      "loss": 0.7791,
       "step": 3720
     },
     {
-      "epoch": 2.0,
-      "learning_rate": 1.4914192318169436e-05,
-      "loss": 3.8524,
+      "epoch": 41.1,
+      "learning_rate": 7.970588235294118e-05,
+      "loss": 0.7931,
       "step": 3740
     },
     {
-      "epoch": 2.0,
-      "step": 3746,
-      "total_flos": 7.191358836520801e+19,
-      "train_loss": 3.6355768978691203,
-      "train_runtime": 71670.0084,
-      "train_samples_per_second": 6.691,
-      "train_steps_per_second": 0.052
+      "epoch": 41.32,
+      "learning_rate": 7.774509803921568e-05,
+      "loss": 0.741,
+      "step": 3760
+    },
+    {
+      "epoch": 41.54,
+      "learning_rate": 7.57843137254902e-05,
+      "loss": 0.7434,
+      "step": 3780
+    },
+    {
+      "epoch": 41.75,
+      "learning_rate": 7.38235294117647e-05,
+      "loss": 0.771,
+      "step": 3800
+    },
+    {
+      "epoch": 41.75,
+      "eval_loss": 0.1530592143535614,
+      "eval_runtime": 186.414,
+      "eval_samples_per_second": 25.98,
+      "eval_steps_per_second": 0.815,
+      "eval_wer": 0.13170593244393983,
+      "step": 3800
+    },
+    {
+      "epoch": 41.97,
+      "learning_rate": 7.186274509803923e-05,
+      "loss": 0.7765,
+      "step": 3820
+    },
+    {
+      "epoch": 42.2,
+      "learning_rate": 6.990196078431373e-05,
+      "loss": 0.7599,
+      "step": 3840
+    },
+    {
+      "epoch": 42.42,
+      "learning_rate": 6.794117647058824e-05,
+      "loss": 0.7782,
+      "step": 3860
+    },
+    {
+      "epoch": 42.63,
+      "learning_rate": 6.598039215686274e-05,
+      "loss": 0.7395,
+      "step": 3880
+    },
+    {
+      "epoch": 42.85,
+      "learning_rate": 6.401960784313726e-05,
+      "loss": 0.7594,
+      "step": 3900
+    },
+    {
+      "epoch": 42.85,
+      "eval_loss": 0.14983513951301575,
+      "eval_runtime": 190.4439,
+      "eval_samples_per_second": 25.43,
+      "eval_steps_per_second": 0.798,
+      "eval_wer": 0.12883905762134545,
+      "step": 3900
+    },
+    {
+      "epoch": 43.08,
+      "learning_rate": 6.205882352941176e-05,
+      "loss": 0.7691,
+      "step": 3920
+    },
+    {
+      "epoch": 43.3,
+      "learning_rate": 6.0098039215686274e-05,
+      "loss": 0.728,
+      "step": 3940
+    },
+    {
+      "epoch": 43.51,
+      "learning_rate": 5.813725490196078e-05,
+      "loss": 0.7585,
+      "step": 3960
+    },
+    {
+      "epoch": 43.73,
+      "learning_rate": 5.6176470588235296e-05,
+      "loss": 0.7564,
+      "step": 3980
+    },
+    {
+      "epoch": 43.95,
+      "learning_rate": 5.4215686274509804e-05,
+      "loss": 0.7383,
+      "step": 4000
+    },
+    {
+      "epoch": 43.95,
+      "eval_loss": 0.15269021689891815,
+      "eval_runtime": 188.5816,
+      "eval_samples_per_second": 25.681,
+      "eval_steps_per_second": 0.806,
+      "eval_wer": 0.13003122338915696,
+      "step": 4000
+    },
+    {
+      "epoch": 44.17,
+      "learning_rate": 5.225490196078431e-05,
+      "loss": 0.7693,
+      "step": 4020
+    },
+    {
+      "epoch": 44.39,
+      "learning_rate": 5.0294117647058826e-05,
+      "loss": 0.7347,
+      "step": 4040
+    },
+    {
+      "epoch": 44.61,
+      "learning_rate": 4.8333333333333334e-05,
+      "loss": 0.7185,
+      "step": 4060
+    },
+    {
+      "epoch": 44.83,
+      "learning_rate": 4.637254901960784e-05,
+      "loss": 0.7394,
+      "step": 4080
+    },
+    {
+      "epoch": 45.05,
+      "learning_rate": 4.4411764705882356e-05,
+      "loss": 0.7565,
+      "step": 4100
+    },
+    {
+      "epoch": 45.05,
+      "eval_loss": 0.14823457598686218,
+      "eval_runtime": 191.1254,
+      "eval_samples_per_second": 25.339,
+      "eval_steps_per_second": 0.795,
+      "eval_wer": 0.12886744252057905,
+      "step": 4100
+    },
+    {
+      "epoch": 45.27,
+      "learning_rate": 4.2450980392156864e-05,
+      "loss": 0.7535,
+      "step": 4120
+    },
+    {
+      "epoch": 45.49,
+      "learning_rate": 4.049019607843137e-05,
+      "loss": 0.737,
+      "step": 4140
+    },
+    {
+      "epoch": 45.71,
+      "learning_rate": 3.852941176470588e-05,
+      "loss": 0.7501,
+      "step": 4160
+    },
+    {
+      "epoch": 45.93,
+      "learning_rate": 3.6568627450980393e-05,
+      "loss": 0.7285,
+      "step": 4180
+    },
+    {
+      "epoch": 46.15,
+      "learning_rate": 3.46078431372549e-05,
+      "loss": 0.7697,
+      "step": 4200
+    },
+    {
+      "epoch": 46.15,
+      "eval_loss": 0.1494804471731186,
+      "eval_runtime": 190.3868,
+      "eval_samples_per_second": 25.438,
+      "eval_steps_per_second": 0.798,
+      "eval_wer": 0.1271927334657962,
+      "step": 4200
+    },
+    {
+      "epoch": 46.37,
+      "learning_rate": 3.264705882352941e-05,
+      "loss": 0.7283,
+      "step": 4220
+    },
+    {
+      "epoch": 46.59,
+      "learning_rate": 3.0686274509803923e-05,
+      "loss": 0.7218,
+      "step": 4240
+    },
+    {
+      "epoch": 46.81,
+      "learning_rate": 2.872549019607843e-05,
+      "loss": 0.7341,
+      "step": 4260
+    },
+    {
+      "epoch": 47.03,
+      "learning_rate": 2.6764705882352942e-05,
+      "loss": 0.7293,
+      "step": 4280
+    },
+    {
+      "epoch": 47.25,
+      "learning_rate": 2.480392156862745e-05,
+      "loss": 0.7194,
+      "step": 4300
+    },
+    {
+      "epoch": 47.25,
+      "eval_loss": 0.14928147196769714,
+      "eval_runtime": 189.625,
+      "eval_samples_per_second": 25.54,
+      "eval_steps_per_second": 0.802,
+      "eval_wer": 0.12693726937269373,
+      "step": 4300
+    },
+    {
+      "epoch": 47.47,
+      "learning_rate": 2.284313725490196e-05,
+      "loss": 0.7326,
+      "step": 4320
+    },
+    {
+      "epoch": 47.69,
+      "learning_rate": 2.088235294117647e-05,
+      "loss": 0.724,
+      "step": 4340
+    },
+    {
+      "epoch": 47.91,
+      "learning_rate": 1.892156862745098e-05,
+      "loss": 0.7316,
+      "step": 4360
+    },
+    {
+      "epoch": 48.13,
+      "learning_rate": 1.696078431372549e-05,
+      "loss": 0.7566,
+      "step": 4380
+    },
+    {
+      "epoch": 48.35,
+      "learning_rate": 1.5e-05,
+      "loss": 0.7479,
+      "step": 4400
+    },
+    {
+      "epoch": 48.35,
+      "eval_loss": 0.1490115076303482,
+      "eval_runtime": 188.3681,
+      "eval_samples_per_second": 25.71,
+      "eval_steps_per_second": 0.807,
+      "eval_wer": 0.12761850695430033,
+      "step": 4400
+    },
+    {
+      "epoch": 48.57,
+      "learning_rate": 1.303921568627451e-05,
+      "loss": 0.7139,
+      "step": 4420
+    },
+    {
+      "epoch": 48.79,
+      "learning_rate": 1.107843137254902e-05,
+      "loss": 0.7345,
+      "step": 4440
+    },
+    {
+      "epoch": 49.01,
+      "learning_rate": 9.117647058823529e-06,
+      "loss": 0.756,
+      "step": 4460
+    },
+    {
+      "epoch": 49.23,
+      "learning_rate": 7.15686274509804e-06,
+      "loss": 0.7196,
+      "step": 4480
+    },
+    {
+      "epoch": 49.45,
+      "learning_rate": 5.196078431372549e-06,
+      "loss": 0.7132,
+      "step": 4500
+    },
+    {
+      "epoch": 49.45,
+      "eval_loss": 0.1501067876815796,
+      "eval_runtime": 190.1989,
+      "eval_samples_per_second": 25.463,
+      "eval_steps_per_second": 0.799,
+      "eval_wer": 0.1265398807834232,
+      "step": 4500
+    },
+    {
+      "epoch": 49.67,
+      "learning_rate": 3.235294117647059e-06,
+      "loss": 0.7278,
+      "step": 4520
+    },
+    {
+      "epoch": 49.89,
+      "learning_rate": 1.2745098039215688e-06,
+      "loss": 0.7431,
+      "step": 4540
+    },
+    {
+      "epoch": 49.99,
+      "step": 4550,
+      "total_flos": 7.03270196503015e+19,
+      "train_loss": 1.3462627654023223,
+      "train_runtime": 35973.7,
+      "train_samples_per_second": 16.248,
+      "train_steps_per_second": 0.126
     }
   ],
-  "max_steps": 3746,
-  "num_train_epochs": 2,
-  "total_flos": 7.191358836520801e+19,
+  "max_steps": 4550,
+  "num_train_epochs": 50,
+  "total_flos": 7.03270196503015e+19,
   "trial_name": null,
   "trial_params": null
 }