End of training

README.md

@@ -15,7 +15,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 # whisper-large-v3-turbo-ft-cv-cy-train-all-plus-other-with-excluded
 
-This model is a fine-tuned version of [openai/whisper-large-v3-turbo](https://huggingface.co/openai/whisper-large-v3-turbo) on an unknown dataset.
+This model is a fine-tuned version of [openai/whisper-large-v3-turbo](https://huggingface.co/openai/whisper-large-v3-turbo) on the DewiBrynJones/commonvoice_18_0_cy train_all+other_with_excluded main dataset.
 It achieves the following results on the evaluation set:
 - Loss: 0.3134
 - Wer: 0.1746

all_results.json

@@ -1,15 +1,15 @@
 {
     "epoch": 7.072135785007072,
-    "eval_loss": 0.3181316554546356,
-    "eval_runtime": 2245.0023,
+    "eval_loss": 0.3134021461009979,
+    "eval_runtime": 1754.3748,
     "eval_samples": 5381,
-    "eval_samples_per_second": 2.397,
-    "eval_steps_per_second": 0.15,
-    "eval_wer": 0.16825718716449958,
-    "total_flos": 5.4329055670370304e+20,
-    "train_loss": 0.09240171913430095,
-    "train_runtime": 60495.4593,
+    "eval_samples_per_second": 3.067,
+    "eval_steps_per_second": 0.192,
+    "eval_wer": 0.17463915066205415,
+    "total_flos": 2.7263703386750976e+20,
+    "train_loss": 0.12853609090298415,
+    "train_runtime": 49407.2097,
     "train_samples": 22611,
-    "train_samples_per_second": 2.645,
-    "train_steps_per_second": 0.083
+    "train_samples_per_second": 3.238,
+    "train_steps_per_second": 0.101
 }

eval_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 7.072135785007072,
-    "eval_loss": 0.3181316554546356,
-    "eval_runtime": 2245.0023,
+    "eval_loss": 0.3134021461009979,
+    "eval_runtime": 1754.3748,
     "eval_samples": 5381,
-    "eval_samples_per_second": 2.397,
-    "eval_steps_per_second": 0.15,
-    "eval_wer": 0.16825718716449958
+    "eval_samples_per_second": 3.067,
+    "eval_steps_per_second": 0.192,
+    "eval_wer": 0.17463915066205415
 }

runs/Oct10_16-21-13_4fc0a42a8f03/events.out.tfevents.1728625433.4fc0a42a8f03.181.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fad36cb131dd8c4c25299af26b298f31c10490b3d16e2caef22919480f20bee5
+size 406

train_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 7.072135785007072,
-    "total_flos": 5.4329055670370304e+20,
-    "train_loss": 0.09240171913430095,
-    "train_runtime": 60495.4593,
+    "total_flos": 2.7263703386750976e+20,
+    "train_loss": 0.12853609090298415,
+    "train_runtime": 49407.2097,
     "train_samples": 22611,
-    "train_samples_per_second": 2.645,
-    "train_steps_per_second": 0.083
+    "train_samples_per_second": 3.238,
+    "train_steps_per_second": 0.101
 }

trainer_state.json

@@ -10,1457 +10,1457 @@
   "log_history": [
     {
       "epoch": 0.03536067892503536,
-      "grad_norm": 5.833098411560059,
-      "learning_rate": 5.000000000000001e-07,
-      "loss": 1.0626,
+      "grad_norm": 15.248167037963867,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 1.6047,
       "step": 25
     },
     {
       "epoch": 0.07072135785007072,
-      "grad_norm": 5.129093647003174,
-      "learning_rate": 1.0000000000000002e-06,
-      "loss": 0.7439,
+      "grad_norm": 11.99557876586914,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 0.7922,
       "step": 50
     },
     {
       "epoch": 0.10608203677510608,
-      "grad_norm": 4.212141513824463,
-      "learning_rate": 1.5e-06,
-      "loss": 0.5203,
+      "grad_norm": 10.12234878540039,
+      "learning_rate": 1.48e-06,
+      "loss": 0.6289,
       "step": 75
     },
     {
       "epoch": 0.14144271570014144,
-      "grad_norm": 3.9624226093292236,
-      "learning_rate": 2.0000000000000003e-06,
-      "loss": 0.4623,
+      "grad_norm": 10.784132957458496,
+      "learning_rate": 1.98e-06,
+      "loss": 0.5627,
       "step": 100
     },
     {
       "epoch": 0.1768033946251768,
-      "grad_norm": 4.4477338790893555,
-      "learning_rate": 2.5e-06,
-      "loss": 0.4498,
+      "grad_norm": 10.604392051696777,
+      "learning_rate": 2.4800000000000004e-06,
+      "loss": 0.563,
       "step": 125
     },
     {
       "epoch": 0.21216407355021216,
-      "grad_norm": 3.65051007270813,
-      "learning_rate": 3e-06,
-      "loss": 0.4272,
+      "grad_norm": 9.61458969116211,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 0.5218,
       "step": 150
     },
     {
       "epoch": 0.24752475247524752,
-      "grad_norm": 4.097436904907227,
-      "learning_rate": 3.5e-06,
-      "loss": 0.4245,
+      "grad_norm": 10.1797513961792,
+      "learning_rate": 3.48e-06,
+      "loss": 0.5316,
       "step": 175
     },
     {
       "epoch": 0.2828854314002829,
-      "grad_norm": 3.8615634441375732,
-      "learning_rate": 4.000000000000001e-06,
-      "loss": 0.4303,
+      "grad_norm": 9.834571838378906,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 0.5232,
       "step": 200
     },
     {
       "epoch": 0.31824611032531824,
-      "grad_norm": 3.634308099746704,
-      "learning_rate": 4.5e-06,
-      "loss": 0.3885,
+      "grad_norm": 9.805900573730469,
+      "learning_rate": 4.48e-06,
+      "loss": 0.4902,
       "step": 225
     },
     {
       "epoch": 0.3536067892503536,
-      "grad_norm": 3.808622360229492,
-      "learning_rate": 5e-06,
-      "loss": 0.3891,
+      "grad_norm": 10.328847885131836,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 0.4885,
       "step": 250
     },
     {
       "epoch": 0.38896746817538896,
-      "grad_norm": 3.767564058303833,
-      "learning_rate": 5.500000000000001e-06,
-      "loss": 0.3737,
+      "grad_norm": 9.624419212341309,
+      "learning_rate": 5.480000000000001e-06,
+      "loss": 0.4615,
       "step": 275
     },
     {
       "epoch": 0.4243281471004243,
-      "grad_norm": 4.20762825012207,
-      "learning_rate": 6e-06,
-      "loss": 0.3549,
+      "grad_norm": 12.492682456970215,
+      "learning_rate": 5.98e-06,
+      "loss": 0.4366,
       "step": 300
     },
     {
       "epoch": 0.4596888260254597,
-      "grad_norm": 3.8779282569885254,
-      "learning_rate": 6.5000000000000004e-06,
-      "loss": 0.3757,
+      "grad_norm": 9.505845069885254,
+      "learning_rate": 6.480000000000001e-06,
+      "loss": 0.4758,
       "step": 325
     },
     {
       "epoch": 0.49504950495049505,
-      "grad_norm": 3.9535462856292725,
-      "learning_rate": 7e-06,
-      "loss": 0.3454,
+      "grad_norm": 8.755816459655762,
+      "learning_rate": 6.98e-06,
+      "loss": 0.44,
       "step": 350
     },
     {
       "epoch": 0.5304101838755304,
-      "grad_norm": 3.7318906784057617,
-      "learning_rate": 7.500000000000001e-06,
-      "loss": 0.3216,
+      "grad_norm": 9.016213417053223,
+      "learning_rate": 7.48e-06,
+      "loss": 0.4061,
       "step": 375
     },
     {
       "epoch": 0.5657708628005658,
-      "grad_norm": 4.168917655944824,
-      "learning_rate": 8.000000000000001e-06,
-      "loss": 0.3194,
+      "grad_norm": 9.07496166229248,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 0.4171,
       "step": 400
     },
     {
       "epoch": 0.6011315417256011,
-      "grad_norm": 3.1413733959198,
-      "learning_rate": 8.5e-06,
-      "loss": 0.3188,
+      "grad_norm": 7.877910137176514,
+      "learning_rate": 8.48e-06,
+      "loss": 0.4141,
       "step": 425
     },
     {
       "epoch": 0.6364922206506365,
-      "grad_norm": 3.511763334274292,
-      "learning_rate": 9e-06,
-      "loss": 0.3129,
+      "grad_norm": 8.071303367614746,
+      "learning_rate": 8.98e-06,
+      "loss": 0.3969,
       "step": 450
     },
     {
       "epoch": 0.6718528995756718,
-      "grad_norm": 3.362168073654175,
-      "learning_rate": 9.5e-06,
-      "loss": 0.3028,
+      "grad_norm": 8.29423713684082,
+      "learning_rate": 9.48e-06,
+      "loss": 0.3924,
       "step": 475
     },
     {
       "epoch": 0.7072135785007072,
-      "grad_norm": 3.6675732135772705,
-      "learning_rate": 1e-05,
-      "loss": 0.3189,
+      "grad_norm": 8.465646743774414,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.4029,
       "step": 500
     },
     {
       "epoch": 0.7425742574257426,
-      "grad_norm": 3.5820181369781494,
-      "learning_rate": 9.944444444444445e-06,
-      "loss": 0.297,
+      "grad_norm": 7.924498081207275,
+      "learning_rate": 9.946666666666667e-06,
+      "loss": 0.379,
       "step": 525
     },
     {
       "epoch": 0.7779349363507779,
-      "grad_norm": 3.423774003982544,
-      "learning_rate": 9.88888888888889e-06,
-      "loss": 0.2935,
+      "grad_norm": 8.063438415527344,
+      "learning_rate": 9.891111111111113e-06,
+      "loss": 0.3814,
       "step": 550
     },
     {
       "epoch": 0.8132956152758133,
-      "grad_norm": 3.7895655632019043,
-      "learning_rate": 9.833333333333333e-06,
-      "loss": 0.2718,
+      "grad_norm": 7.638169765472412,
+      "learning_rate": 9.835555555555556e-06,
+      "loss": 0.3476,
       "step": 575
     },
     {
       "epoch": 0.8486562942008486,
-      "grad_norm": 3.2421023845672607,
-      "learning_rate": 9.777777777777779e-06,
-      "loss": 0.2741,
+      "grad_norm": 8.318132400512695,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 0.3499,
       "step": 600
     },
     {
       "epoch": 0.884016973125884,
-      "grad_norm": 3.0554616451263428,
-      "learning_rate": 9.722222222222223e-06,
-      "loss": 0.2661,
+      "grad_norm": 7.153702735900879,
+      "learning_rate": 9.724444444444445e-06,
+      "loss": 0.3411,
       "step": 625
     },
     {
       "epoch": 0.9193776520509194,
-      "grad_norm": 2.754528760910034,
-      "learning_rate": 9.666666666666667e-06,
-      "loss": 0.2573,
+      "grad_norm": 5.998073101043701,
+      "learning_rate": 9.671111111111112e-06,
+      "loss": 0.3206,
       "step": 650
     },
     {
       "epoch": 0.9547383309759547,
-      "grad_norm": 2.9093120098114014,
-      "learning_rate": 9.611111111111112e-06,
-      "loss": 0.2766,
+      "grad_norm": 7.312229156494141,
+      "learning_rate": 9.615555555555558e-06,
+      "loss": 0.3463,
       "step": 675
     },
     {
       "epoch": 0.9900990099009901,
-      "grad_norm": 2.9693257808685303,
-      "learning_rate": 9.555555555555556e-06,
-      "loss": 0.2506,
+      "grad_norm": 6.737645626068115,
+      "learning_rate": 9.56e-06,
+      "loss": 0.3285,
       "step": 700
     },
     {
       "epoch": 1.0254596888260255,
-      "grad_norm": 2.879349946975708,
-      "learning_rate": 9.5e-06,
-      "loss": 0.2021,
+      "grad_norm": 6.451304912567139,
+      "learning_rate": 9.504444444444446e-06,
+      "loss": 0.2746,
       "step": 725
     },
     {
       "epoch": 1.0608203677510608,
-      "grad_norm": 2.6136014461517334,
-      "learning_rate": 9.444444444444445e-06,
-      "loss": 0.181,
+      "grad_norm": 5.666613578796387,
+      "learning_rate": 9.44888888888889e-06,
+      "loss": 0.2455,
       "step": 750
     },
     {
       "epoch": 1.0961810466760962,
-      "grad_norm": 2.8699588775634766,
-      "learning_rate": 9.38888888888889e-06,
-      "loss": 0.1648,
+      "grad_norm": 6.165246963500977,
+      "learning_rate": 9.393333333333334e-06,
+      "loss": 0.2246,
       "step": 775
     },
     {
       "epoch": 1.1315417256011315,
-      "grad_norm": 2.2577948570251465,
-      "learning_rate": 9.333333333333334e-06,
-      "loss": 0.171,
+      "grad_norm": 5.314395904541016,
+      "learning_rate": 9.33777777777778e-06,
+      "loss": 0.2313,
       "step": 800
     },
     {
       "epoch": 1.166902404526167,
-      "grad_norm": 2.6615614891052246,
-      "learning_rate": 9.277777777777778e-06,
-      "loss": 0.1759,
+      "grad_norm": 6.031562328338623,
+      "learning_rate": 9.282222222222222e-06,
+      "loss": 0.2445,
       "step": 825
     },
     {
       "epoch": 1.2022630834512023,
-      "grad_norm": 3.6334331035614014,
-      "learning_rate": 9.222222222222224e-06,
-      "loss": 0.1703,
+      "grad_norm": 8.143284797668457,
+      "learning_rate": 9.226666666666668e-06,
+      "loss": 0.2248,
       "step": 850
     },
     {
       "epoch": 1.2376237623762376,
-      "grad_norm": 2.0051722526550293,
-      "learning_rate": 9.166666666666666e-06,
-      "loss": 0.1666,
+      "grad_norm": 4.90247106552124,
+      "learning_rate": 9.171111111111112e-06,
+      "loss": 0.2296,
       "step": 875
     },
     {
       "epoch": 1.272984441301273,
-      "grad_norm": 2.5868630409240723,
-      "learning_rate": 9.111111111111112e-06,
-      "loss": 0.1631,
+      "grad_norm": 5.334079265594482,
+      "learning_rate": 9.115555555555556e-06,
+      "loss": 0.2252,
       "step": 900
     },
     {
       "epoch": 1.3083451202263083,
-      "grad_norm": 2.7406225204467773,
-      "learning_rate": 9.055555555555556e-06,
-      "loss": 0.1625,
+      "grad_norm": 6.497454643249512,
+      "learning_rate": 9.060000000000001e-06,
+      "loss": 0.2261,
       "step": 925
     },
     {
       "epoch": 1.3437057991513437,
-      "grad_norm": 2.9130780696868896,
-      "learning_rate": 9e-06,
-      "loss": 0.1779,
+      "grad_norm": 7.067910671234131,
+      "learning_rate": 9.004444444444445e-06,
+      "loss": 0.2416,
       "step": 950
     },
     {
       "epoch": 1.379066478076379,
-      "grad_norm": 2.973248243331909,
-      "learning_rate": 8.944444444444446e-06,
-      "loss": 0.1642,
+      "grad_norm": 8.470537185668945,
+      "learning_rate": 8.94888888888889e-06,
+      "loss": 0.2219,
       "step": 975
     },
     {
       "epoch": 1.4144271570014144,
-      "grad_norm": 2.1427431106567383,
-      "learning_rate": 8.888888888888888e-06,
-      "loss": 0.1599,
+      "grad_norm": 4.9936747550964355,
+      "learning_rate": 8.893333333333333e-06,
+      "loss": 0.2147,
       "step": 1000
     },
     {
       "epoch": 1.4144271570014144,
-      "eval_loss": 0.2516109049320221,
-      "eval_runtime": 2179.6002,
-      "eval_samples_per_second": 2.469,
-      "eval_steps_per_second": 0.155,
-      "eval_wer": 0.2035468607101674,
+      "eval_loss": 0.3066270351409912,
+      "eval_runtime": 1765.706,
+      "eval_samples_per_second": 3.048,
+      "eval_steps_per_second": 0.191,
+      "eval_wer": 0.23493975903614459,
       "step": 1000
     },
     {
       "epoch": 1.4497878359264498,
-      "grad_norm": 3.01638126373291,
-      "learning_rate": 8.833333333333334e-06,
-      "loss": 0.1525,
+      "grad_norm": 7.876005172729492,
+      "learning_rate": 8.83777777777778e-06,
+      "loss": 0.2032,
       "step": 1025
     },
     {
       "epoch": 1.4851485148514851,
-      "grad_norm": 2.6232805252075195,
-      "learning_rate": 8.777777777777778e-06,
-      "loss": 0.1565,
+      "grad_norm": 6.1690826416015625,
+      "learning_rate": 8.782222222222223e-06,
+      "loss": 0.216,
       "step": 1050
     },
     {
       "epoch": 1.5205091937765205,
-      "grad_norm": 2.569593906402588,
-      "learning_rate": 8.722222222222224e-06,
-      "loss": 0.1546,
+      "grad_norm": 5.094773769378662,
+      "learning_rate": 8.726666666666667e-06,
+      "loss": 0.2189,
       "step": 1075
     },
     {
       "epoch": 1.5558698727015559,
-      "grad_norm": 2.730534076690674,
-      "learning_rate": 8.666666666666668e-06,
-      "loss": 0.1523,
+      "grad_norm": 6.5309834480285645,
+      "learning_rate": 8.671111111111113e-06,
+      "loss": 0.2188,
       "step": 1100
     },
     {
       "epoch": 1.5912305516265912,
-      "grad_norm": 3.71217942237854,
-      "learning_rate": 8.611111111111112e-06,
-      "loss": 0.1601,
+      "grad_norm": 7.060642242431641,
+      "learning_rate": 8.615555555555555e-06,
+      "loss": 0.2233,
       "step": 1125
     },
     {
       "epoch": 1.6265912305516266,
-      "grad_norm": 2.2378976345062256,
-      "learning_rate": 8.555555555555556e-06,
-      "loss": 0.1517,
+      "grad_norm": 5.661276817321777,
+      "learning_rate": 8.560000000000001e-06,
+      "loss": 0.2124,
       "step": 1150
     },
     {
       "epoch": 1.661951909476662,
-      "grad_norm": 2.2248027324676514,
-      "learning_rate": 8.5e-06,
-      "loss": 0.1479,
+      "grad_norm": 5.289186477661133,
+      "learning_rate": 8.504444444444445e-06,
+      "loss": 0.2022,
       "step": 1175
     },
     {
       "epoch": 1.6973125884016973,
-      "grad_norm": 1.97626531124115,
-      "learning_rate": 8.444444444444446e-06,
-      "loss": 0.1448,
+      "grad_norm": 4.7772064208984375,
+      "learning_rate": 8.448888888888889e-06,
+      "loss": 0.2018,
       "step": 1200
     },
     {
       "epoch": 1.7326732673267327,
-      "grad_norm": 2.5063023567199707,
-      "learning_rate": 8.38888888888889e-06,
-      "loss": 0.1493,
+      "grad_norm": 5.6891303062438965,
+      "learning_rate": 8.393333333333335e-06,
+      "loss": 0.2059,
       "step": 1225
     },
     {
       "epoch": 1.768033946251768,
-      "grad_norm": 2.230543613433838,
-      "learning_rate": 8.333333333333334e-06,
-      "loss": 0.1413,
+      "grad_norm": 5.304312229156494,
+      "learning_rate": 8.337777777777777e-06,
+      "loss": 0.1949,
       "step": 1250
     },
     {
       "epoch": 1.8033946251768034,
-      "grad_norm": 2.3696229457855225,
-      "learning_rate": 8.277777777777778e-06,
-      "loss": 0.1514,
+      "grad_norm": 5.472725868225098,
+      "learning_rate": 8.282222222222223e-06,
+      "loss": 0.21,
       "step": 1275
     },
     {
       "epoch": 1.8387553041018387,
-      "grad_norm": 2.3994140625,
-      "learning_rate": 8.222222222222222e-06,
-      "loss": 0.1514,
+      "grad_norm": 4.937655448913574,
+      "learning_rate": 8.226666666666667e-06,
+      "loss": 0.2137,
       "step": 1300
     },
     {
       "epoch": 1.874115983026874,
-      "grad_norm": 2.1594157218933105,
-      "learning_rate": 8.166666666666668e-06,
-      "loss": 0.1391,
+      "grad_norm": 6.363350868225098,
+      "learning_rate": 8.171111111111113e-06,
+      "loss": 0.1919,
       "step": 1325
     },
     {
       "epoch": 1.9094766619519095,
-      "grad_norm": 2.2677788734436035,
-      "learning_rate": 8.111111111111112e-06,
-      "loss": 0.1333,
+      "grad_norm": 5.087652206420898,
+      "learning_rate": 8.115555555555557e-06,
+      "loss": 0.1783,
       "step": 1350
     },
     {
       "epoch": 1.9448373408769448,
-      "grad_norm": 2.0969929695129395,
-      "learning_rate": 8.055555555555557e-06,
-      "loss": 0.141,
+      "grad_norm": 4.039090633392334,
+      "learning_rate": 8.06e-06,
+      "loss": 0.1879,
       "step": 1375
     },
     {
       "epoch": 1.9801980198019802,
-      "grad_norm": 2.2506678104400635,
-      "learning_rate": 8.000000000000001e-06,
-      "loss": 0.1393,
+      "grad_norm": 5.647103309631348,
+      "learning_rate": 8.004444444444445e-06,
+      "loss": 0.1904,
       "step": 1400
     },
     {
       "epoch": 2.0155586987270158,
-      "grad_norm": 1.4057811498641968,
-      "learning_rate": 7.944444444444445e-06,
-      "loss": 0.1088,
+      "grad_norm": 3.8179798126220703,
+      "learning_rate": 7.948888888888889e-06,
+      "loss": 0.1572,
       "step": 1425
     },
     {
       "epoch": 2.050919377652051,
-      "grad_norm": 1.8405675888061523,
-      "learning_rate": 7.88888888888889e-06,
-      "loss": 0.0696,
+      "grad_norm": 5.048496246337891,
+      "learning_rate": 7.893333333333335e-06,
+      "loss": 0.1087,
       "step": 1450
     },
     {
       "epoch": 2.0862800565770865,
-      "grad_norm": 1.7965513467788696,
-      "learning_rate": 7.833333333333333e-06,
-      "loss": 0.0715,
+      "grad_norm": 3.450915813446045,
+      "learning_rate": 7.837777777777779e-06,
+      "loss": 0.1088,
       "step": 1475
     },
     {
       "epoch": 2.1216407355021216,
-      "grad_norm": 1.533977746963501,
-      "learning_rate": 7.77777777777778e-06,
-      "loss": 0.0655,
+      "grad_norm": 4.442863464355469,
+      "learning_rate": 7.782222222222223e-06,
+      "loss": 0.1049,
       "step": 1500
     },
     {
       "epoch": 2.157001414427157,
-      "grad_norm": 1.5952216386795044,
-      "learning_rate": 7.722222222222223e-06,
-      "loss": 0.0701,
+      "grad_norm": 4.130264759063721,
+      "learning_rate": 7.726666666666667e-06,
+      "loss": 0.1118,
       "step": 1525
     },
     {
       "epoch": 2.1923620933521923,
-      "grad_norm": 1.796464443206787,
-      "learning_rate": 7.666666666666667e-06,
-      "loss": 0.0649,
+      "grad_norm": 3.8330652713775635,
+      "learning_rate": 7.67111111111111e-06,
+      "loss": 0.1068,
       "step": 1550
     },
     {
       "epoch": 2.227722772277228,
-      "grad_norm": 1.637083888053894,
-      "learning_rate": 7.611111111111111e-06,
-      "loss": 0.0729,
+      "grad_norm": 4.3345489501953125,
+      "learning_rate": 7.6155555555555564e-06,
+      "loss": 0.1133,
       "step": 1575
     },
     {
       "epoch": 2.263083451202263,
-      "grad_norm": 1.361791968345642,
-      "learning_rate": 7.555555555555556e-06,
-      "loss": 0.0661,
+      "grad_norm": 4.276752948760986,
+      "learning_rate": 7.5600000000000005e-06,
+      "loss": 0.1105,
       "step": 1600
     },
     {
       "epoch": 2.298444130127298,
-      "grad_norm": 1.4004043340682983,
-      "learning_rate": 7.500000000000001e-06,
-      "loss": 0.0635,
+      "grad_norm": 3.8332560062408447,
+      "learning_rate": 7.504444444444445e-06,
+      "loss": 0.1011,
       "step": 1625
     },
     {
       "epoch": 2.333804809052334,
-      "grad_norm": 1.415955662727356,
-      "learning_rate": 7.444444444444445e-06,
-      "loss": 0.0694,
+      "grad_norm": 4.976287841796875,
+      "learning_rate": 7.44888888888889e-06,
+      "loss": 0.1151,
       "step": 1650
     },
     {
       "epoch": 2.3691654879773694,
-      "grad_norm": 1.5830997228622437,
-      "learning_rate": 7.38888888888889e-06,
-      "loss": 0.0693,
+      "grad_norm": 3.8312933444976807,
+      "learning_rate": 7.393333333333333e-06,
+      "loss": 0.1124,
       "step": 1675
     },
     {
       "epoch": 2.4045261669024045,
-      "grad_norm": 1.8817075490951538,
-      "learning_rate": 7.333333333333333e-06,
-      "loss": 0.0753,
+      "grad_norm": 4.4323248863220215,
+      "learning_rate": 7.337777777777778e-06,
+      "loss": 0.1148,
       "step": 1700
     },
     {
       "epoch": 2.4398868458274396,
-      "grad_norm": 2.290759563446045,
-      "learning_rate": 7.277777777777778e-06,
-      "loss": 0.0705,
+      "grad_norm": 3.492539882659912,
+      "learning_rate": 7.282222222222222e-06,
+      "loss": 0.112,
       "step": 1725
     },
     {
       "epoch": 2.4752475247524752,
-      "grad_norm": 1.521173119544983,
-      "learning_rate": 7.222222222222223e-06,
-      "loss": 0.0745,
+      "grad_norm": 4.846162796020508,
+      "learning_rate": 7.226666666666667e-06,
+      "loss": 0.1128,
       "step": 1750
     },
     {
       "epoch": 2.510608203677511,
-      "grad_norm": 2.191132068634033,
-      "learning_rate": 7.166666666666667e-06,
-      "loss": 0.0678,
+      "grad_norm": 5.806790828704834,
+      "learning_rate": 7.171111111111112e-06,
+      "loss": 0.1072,
       "step": 1775
     },
     {
       "epoch": 2.545968882602546,
-      "grad_norm": 1.374758243560791,
-      "learning_rate": 7.111111111111112e-06,
-      "loss": 0.0622,
+      "grad_norm": 3.3074755668640137,
+      "learning_rate": 7.115555555555557e-06,
+      "loss": 0.1067,
       "step": 1800
     },
     {
       "epoch": 2.581329561527581,
-      "grad_norm": 1.8768402338027954,
-      "learning_rate": 7.055555555555557e-06,
-      "loss": 0.0596,
+      "grad_norm": 3.70501446723938,
+      "learning_rate": 7.06e-06,
+      "loss": 0.1016,
       "step": 1825
     },
     {
       "epoch": 2.6166902404526167,
-      "grad_norm": 1.5529979467391968,
-      "learning_rate": 7e-06,
-      "loss": 0.063,
+      "grad_norm": 2.6926796436309814,
+      "learning_rate": 7.004444444444445e-06,
+      "loss": 0.1018,
       "step": 1850
     },
     {
       "epoch": 2.6520509193776522,
-      "grad_norm": 1.5192201137542725,
-      "learning_rate": 6.944444444444445e-06,
-      "loss": 0.072,
+      "grad_norm": 4.052884578704834,
+      "learning_rate": 6.948888888888889e-06,
+      "loss": 0.1106,
       "step": 1875
     },
     {
       "epoch": 2.6874115983026874,
-      "grad_norm": 1.872154951095581,
-      "learning_rate": 6.88888888888889e-06,
-      "loss": 0.069,
+      "grad_norm": 4.722410678863525,
+      "learning_rate": 6.893333333333334e-06,
+      "loss": 0.1133,
       "step": 1900
     },
     {
       "epoch": 2.7227722772277225,
-      "grad_norm": 1.8019146919250488,
-      "learning_rate": 6.833333333333334e-06,
-      "loss": 0.0599,
+      "grad_norm": 4.194309711456299,
+      "learning_rate": 6.837777777777779e-06,
+      "loss": 0.1012,
       "step": 1925
     },
     {
       "epoch": 2.758132956152758,
-      "grad_norm": 1.5129454135894775,
-      "learning_rate": 6.777777777777779e-06,
-      "loss": 0.0671,
+      "grad_norm": 3.4879953861236572,
+      "learning_rate": 6.782222222222222e-06,
+      "loss": 0.1091,
       "step": 1950
     },
     {
       "epoch": 2.7934936350777937,
-      "grad_norm": 1.387603998184204,
-      "learning_rate": 6.7222222222222235e-06,
-      "loss": 0.059,
+      "grad_norm": 2.7147629261016846,
+      "learning_rate": 6.726666666666667e-06,
+      "loss": 0.0909,
       "step": 1975
     },
     {
       "epoch": 2.828854314002829,
-      "grad_norm": 1.5892807245254517,
-      "learning_rate": 6.666666666666667e-06,
-      "loss": 0.0651,
+      "grad_norm": 4.304833889007568,
+      "learning_rate": 6.671111111111112e-06,
+      "loss": 0.0989,
       "step": 2000
     },
     {
       "epoch": 2.828854314002829,
-      "eval_loss": 0.23385389149188995,
-      "eval_runtime": 2134.4668,
-      "eval_samples_per_second": 2.521,
-      "eval_steps_per_second": 0.158,
-      "eval_wer": 0.18308879080679152,
+      "eval_loss": 0.2774756848812103,
+      "eval_runtime": 1765.8044,
+      "eval_samples_per_second": 3.047,
+      "eval_steps_per_second": 0.191,
+      "eval_wer": 0.20720505785518312,
       "step": 2000
     },
     {
       "epoch": 2.864214992927864,
-      "grad_norm": 2.636842727661133,
-      "learning_rate": 6.6111111111111115e-06,
-      "loss": 0.0683,
+      "grad_norm": 5.535168170928955,
+      "learning_rate": 6.615555555555556e-06,
+      "loss": 0.1188,
       "step": 2025
     },
     {
       "epoch": 2.8995756718528995,
-      "grad_norm": 2.534872531890869,
-      "learning_rate": 6.555555555555556e-06,
-      "loss": 0.0617,
+      "grad_norm": 5.256195068359375,
+      "learning_rate": 6.560000000000001e-06,
+      "loss": 0.097,
       "step": 2050
     },
     {
       "epoch": 2.934936350777935,
-      "grad_norm": 1.9003880023956299,
-      "learning_rate": 6.5000000000000004e-06,
-      "loss": 0.0566,
+      "grad_norm": 4.056241035461426,
+      "learning_rate": 6.504444444444446e-06,
+      "loss": 0.096,
       "step": 2075
     },
     {
       "epoch": 2.9702970297029703,
-      "grad_norm": 2.0285208225250244,
-      "learning_rate": 6.444444444444445e-06,
-      "loss": 0.0652,
+      "grad_norm": 4.049830913543701,
+      "learning_rate": 6.448888888888889e-06,
+      "loss": 0.1031,
       "step": 2100
     },
     {
       "epoch": 3.005657708628006,
-      "grad_norm": 1.0410995483398438,
-      "learning_rate": 6.3888888888888885e-06,
-      "loss": 0.0567,
+      "grad_norm": 2.4076595306396484,
+      "learning_rate": 6.393333333333334e-06,
+      "loss": 0.0931,
       "step": 2125
     },
     {
       "epoch": 3.041018387553041,
-      "grad_norm": 1.090918779373169,
-      "learning_rate": 6.333333333333333e-06,
-      "loss": 0.0304,
+      "grad_norm": 1.8493136167526245,
+      "learning_rate": 6.3377777777777786e-06,
+      "loss": 0.0591,
       "step": 2150
     },
     {
       "epoch": 3.0763790664780766,
-      "grad_norm": 1.4816592931747437,
-      "learning_rate": 6.277777777777778e-06,
-      "loss": 0.0255,
+      "grad_norm": 2.923140525817871,
+      "learning_rate": 6.282222222222223e-06,
+      "loss": 0.0498,
       "step": 2175
     },
     {
       "epoch": 3.1117397454031117,
-      "grad_norm": 1.5663459300994873,
-      "learning_rate": 6.222222222222223e-06,
-      "loss": 0.0279,
+      "grad_norm": 3.3305506706237793,
+      "learning_rate": 6.2266666666666675e-06,
+      "loss": 0.0538,
       "step": 2200
     },
     {
       "epoch": 3.1471004243281473,
-      "grad_norm": 1.0855896472930908,
-      "learning_rate": 6.166666666666667e-06,
-      "loss": 0.0257,
+      "grad_norm": 2.396880626678467,
+      "learning_rate": 6.171111111111112e-06,
+      "loss": 0.0496,
       "step": 2225
     },
     {
       "epoch": 3.1824611032531824,
-      "grad_norm": 0.9822140336036682,
-      "learning_rate": 6.111111111111112e-06,
-      "loss": 0.0274,
+      "grad_norm": 3.4060046672821045,
+      "learning_rate": 6.1155555555555555e-06,
+      "loss": 0.0559,
       "step": 2250
     },
     {
       "epoch": 3.217821782178218,
-      "grad_norm": 1.3822486400604248,
-      "learning_rate": 6.055555555555555e-06,
-      "loss": 0.0295,
+      "grad_norm": 4.2796478271484375,
+      "learning_rate": 6.0600000000000004e-06,
+      "loss": 0.0479,
       "step": 2275
     },
     {
       "epoch": 3.253182461103253,
-      "grad_norm": 0.9994338154792786,
-      "learning_rate": 6e-06,
-      "loss": 0.0287,
+      "grad_norm": 3.096123456954956,
+      "learning_rate": 6.004444444444445e-06,
+      "loss": 0.0547,
       "step": 2300
     },
     {
       "epoch": 3.2885431400282887,
-      "grad_norm": 1.223861813545227,
-      "learning_rate": 5.944444444444445e-06,
-      "loss": 0.0288,
+      "grad_norm": 3.5626742839813232,
+      "learning_rate": 5.948888888888889e-06,
+      "loss": 0.058,
       "step": 2325
     },
     {
       "epoch": 3.323903818953324,
-      "grad_norm": 1.0625354051589966,
-      "learning_rate": 5.88888888888889e-06,
-      "loss": 0.0313,
+      "grad_norm": 3.2983179092407227,
+      "learning_rate": 5.893333333333334e-06,
+      "loss": 0.0586,
       "step": 2350
     },
     {
       "epoch": 3.3592644978783595,
-      "grad_norm": 1.2225134372711182,
-      "learning_rate": 5.833333333333334e-06,
-      "loss": 0.0278,
+      "grad_norm": 3.978823184967041,
+      "learning_rate": 5.837777777777777e-06,
+      "loss": 0.0554,
       "step": 2375
     },
     {
       "epoch": 3.3946251768033946,
-      "grad_norm": 1.0246098041534424,
-      "learning_rate": 5.777777777777778e-06,
-      "loss": 0.0285,
+      "grad_norm": 3.5360333919525146,
+      "learning_rate": 5.782222222222222e-06,
+      "loss": 0.0549,
       "step": 2400
     },
     {
       "epoch": 3.42998585572843,
-      "grad_norm": 1.376129388809204,
-      "learning_rate": 5.722222222222222e-06,
-      "loss": 0.0326,
+      "grad_norm": 3.2348575592041016,
+      "learning_rate": 5.726666666666667e-06,
+      "loss": 0.0609,
       "step": 2425
     },
     {
       "epoch": 3.4653465346534653,
-      "grad_norm": 1.7981650829315186,
-      "learning_rate": 5.666666666666667e-06,
-      "loss": 0.0326,
+      "grad_norm": 4.142767906188965,
+      "learning_rate": 5.671111111111112e-06,
+      "loss": 0.0615,
       "step": 2450
     },
     {
       "epoch": 3.500707213578501,
-      "grad_norm": 0.6855577230453491,
-      "learning_rate": 5.611111111111112e-06,
-      "loss": 0.0272,
+      "grad_norm": 3.2305757999420166,
+      "learning_rate": 5.615555555555556e-06,
+      "loss": 0.0485,
       "step": 2475
     },
     {
       "epoch": 3.536067892503536,
-      "grad_norm": 1.0780831575393677,
-      "learning_rate": 5.555555555555557e-06,
-      "loss": 0.026,
+      "grad_norm": 2.9658920764923096,
+      "learning_rate": 5.560000000000001e-06,
+      "loss": 0.054,
       "step": 2500
     },
     {
       "epoch": 3.571428571428571,
-      "grad_norm": 1.4502952098846436,
-      "learning_rate": 5.500000000000001e-06,
-      "loss": 0.0326,
+      "grad_norm": 3.1340267658233643,
+      "learning_rate": 5.504444444444444e-06,
+      "loss": 0.0633,
       "step": 2525
     },
     {
       "epoch": 3.6067892503536068,
-      "grad_norm": 1.2788467407226562,
-      "learning_rate": 5.444444444444445e-06,
-      "loss": 0.0278,
+      "grad_norm": 2.692047119140625,
+      "learning_rate": 5.448888888888889e-06,
+      "loss": 0.0549,
       "step": 2550
     },
     {
       "epoch": 3.6421499292786423,
-      "grad_norm": 1.4065041542053223,
-      "learning_rate": 5.388888888888889e-06,
-      "loss": 0.028,
+      "grad_norm": 2.9761712551116943,
+      "learning_rate": 5.393333333333334e-06,
+      "loss": 0.0537,
       "step": 2575
     },
     {
       "epoch": 3.6775106082036775,
-      "grad_norm": 0.9822909832000732,
-      "learning_rate": 5.333333333333334e-06,
-      "loss": 0.024,
+      "grad_norm": 2.88071346282959,
+      "learning_rate": 5.337777777777779e-06,
+      "loss": 0.0497,
       "step": 2600
     },
     {
       "epoch": 3.7128712871287126,
-      "grad_norm": 0.9713716506958008,
-      "learning_rate": 5.2777777777777785e-06,
-      "loss": 0.0314,
+      "grad_norm": 3.6829402446746826,
+      "learning_rate": 5.282222222222223e-06,
+      "loss": 0.0609,
       "step": 2625
     },
     {
       "epoch": 3.748231966053748,
-      "grad_norm": 1.5902820825576782,
-      "learning_rate": 5.2222222222222226e-06,
-      "loss": 0.0286,
+      "grad_norm": 4.3834614753723145,
+      "learning_rate": 5.226666666666667e-06,
+      "loss": 0.0567,
       "step": 2650
     },
     {
       "epoch": 3.783592644978784,
-      "grad_norm": 1.065138816833496,
-      "learning_rate": 5.1666666666666675e-06,
-      "loss": 0.0237,
+      "grad_norm": 2.6627509593963623,
+      "learning_rate": 5.171111111111111e-06,
+      "loss": 0.052,
       "step": 2675
     },
     {
       "epoch": 3.818953323903819,
-      "grad_norm": 1.4179657697677612,
-      "learning_rate": 5.1111111111111115e-06,
-      "loss": 0.0265,
+      "grad_norm": 3.6826014518737793,
+      "learning_rate": 5.115555555555556e-06,
+      "loss": 0.0529,
       "step": 2700
     },
     {
       "epoch": 3.854314002828854,
-      "grad_norm": 1.563292384147644,
-      "learning_rate": 5.0555555555555555e-06,
-      "loss": 0.0321,
+      "grad_norm": 3.048292636871338,
+      "learning_rate": 5.060000000000001e-06,
+      "loss": 0.0617,
       "step": 2725
     },
     {
       "epoch": 3.8896746817538896,
-      "grad_norm": 1.1048407554626465,
-      "learning_rate": 5e-06,
-      "loss": 0.0277,
+      "grad_norm": 2.5703887939453125,
+      "learning_rate": 5.004444444444445e-06,
+      "loss": 0.0507,
       "step": 2750
     },
     {
       "epoch": 3.9250353606789252,
-      "grad_norm": 1.585411787033081,
-      "learning_rate": 4.944444444444445e-06,
-      "loss": 0.0309,
+      "grad_norm": 2.8455772399902344,
+      "learning_rate": 4.94888888888889e-06,
+      "loss": 0.0562,
       "step": 2775
     },
     {
       "epoch": 3.9603960396039604,
-      "grad_norm": 1.438583493232727,
-      "learning_rate": 4.888888888888889e-06,
-      "loss": 0.0229,
+      "grad_norm": 2.147115468978882,
+      "learning_rate": 4.893333333333334e-06,
+      "loss": 0.0479,
       "step": 2800
     },
     {
       "epoch": 3.9957567185289955,
-      "grad_norm": 1.6560890674591064,
-      "learning_rate": 4.833333333333333e-06,
-      "loss": 0.0265,
+      "grad_norm": 4.647951126098633,
+      "learning_rate": 4.837777777777778e-06,
+      "loss": 0.0555,
       "step": 2825
     },
     {
       "epoch": 4.0311173974540315,
-      "grad_norm": 1.1240108013153076,
-      "learning_rate": 4.777777777777778e-06,
-      "loss": 0.013,
+      "grad_norm": 1.5333168506622314,
+      "learning_rate": 4.7822222222222226e-06,
+      "loss": 0.0241,
       "step": 2850
     },
     {
       "epoch": 4.066478076379067,
-      "grad_norm": 1.5441868305206299,
-      "learning_rate": 4.722222222222222e-06,
-      "loss": 0.013,
+      "grad_norm": 2.611342668533325,
+      "learning_rate": 4.728888888888889e-06,
+      "loss": 0.0297,
       "step": 2875
     },
     {
       "epoch": 4.101838755304102,
-      "grad_norm": 0.6906932592391968,
-      "learning_rate": 4.666666666666667e-06,
-      "loss": 0.0137,
+      "grad_norm": 1.1682252883911133,
+      "learning_rate": 4.673333333333333e-06,
+      "loss": 0.028,
       "step": 2900
     },
     {
       "epoch": 4.137199434229137,
-      "grad_norm": 0.4889713227748871,
-      "learning_rate": 4.611111111111112e-06,
-      "loss": 0.0107,
+      "grad_norm": 1.971526861190796,
+      "learning_rate": 4.617777777777778e-06,
+      "loss": 0.0256,
       "step": 2925
     },
     {
       "epoch": 4.172560113154173,
-      "grad_norm": 1.0169159173965454,
-      "learning_rate": 4.555555555555556e-06,
-      "loss": 0.0105,
+      "grad_norm": 2.9451370239257812,
+      "learning_rate": 4.562222222222222e-06,
+      "loss": 0.0248,
       "step": 2950
     },
     {
       "epoch": 4.207920792079208,
-      "grad_norm": 0.5549736022949219,
-      "learning_rate": 4.5e-06,
-      "loss": 0.0111,
+      "grad_norm": 2.203016757965088,
+      "learning_rate": 4.506666666666667e-06,
+      "loss": 0.0254,
       "step": 2975
     },
     {
       "epoch": 4.243281471004243,
-      "grad_norm": 1.1406062841415405,
-      "learning_rate": 4.444444444444444e-06,
-      "loss": 0.0119,
+      "grad_norm": 3.0037150382995605,
+      "learning_rate": 4.451111111111112e-06,
+      "loss": 0.0295,
       "step": 3000
     },
     {
       "epoch": 4.243281471004243,
-      "eval_loss": 0.2719973921775818,
-      "eval_runtime": 2233.3625,
-      "eval_samples_per_second": 2.409,
-      "eval_steps_per_second": 0.151,
-      "eval_wer": 0.1750566622927353,
+      "eval_loss": 0.29346707463264465,
+      "eval_runtime": 1760.8686,
+      "eval_samples_per_second": 3.056,
+      "eval_steps_per_second": 0.191,
+      "eval_wer": 0.1918962980635413,
       "step": 3000
     },
     {
       "epoch": 4.278642149929278,
-      "grad_norm": 0.460374653339386,
-      "learning_rate": 4.388888888888889e-06,
-      "loss": 0.0085,
+      "grad_norm": 2.936363697052002,
+      "learning_rate": 4.395555555555556e-06,
+      "loss": 0.0235,
       "step": 3025
     },
     {
       "epoch": 4.314002828854314,
-      "grad_norm": 0.334330677986145,
-      "learning_rate": 4.333333333333334e-06,
-      "loss": 0.01,
+      "grad_norm": 1.909324288368225,
+      "learning_rate": 4.34e-06,
+      "loss": 0.027,
       "step": 3050
     },
     {
       "epoch": 4.3493635077793495,
-      "grad_norm": 0.8199196457862854,
-      "learning_rate": 4.277777777777778e-06,
-      "loss": 0.0105,
+      "grad_norm": 1.8977993726730347,
+      "learning_rate": 4.284444444444445e-06,
+      "loss": 0.0267,
       "step": 3075
     },
     {
       "epoch": 4.384724186704385,
-      "grad_norm": 0.9259428977966309,
-      "learning_rate": 4.222222222222223e-06,
-      "loss": 0.0108,
+      "grad_norm": 2.092729330062866,
+      "learning_rate": 4.228888888888889e-06,
+      "loss": 0.0277,
       "step": 3100
     },
     {
       "epoch": 4.42008486562942,
-      "grad_norm": 1.7561092376708984,
-      "learning_rate": 4.166666666666667e-06,
-      "loss": 0.0098,
+      "grad_norm": 2.539109945297241,
+      "learning_rate": 4.173333333333334e-06,
+      "loss": 0.0232,
       "step": 3125
     },
     {
       "epoch": 4.455445544554456,
-      "grad_norm": 0.5159019827842712,
-      "learning_rate": 4.111111111111111e-06,
-      "loss": 0.0114,
+      "grad_norm": 1.8122098445892334,
+      "learning_rate": 4.117777777777779e-06,
+      "loss": 0.0276,
       "step": 3150
     },
     {
       "epoch": 4.490806223479491,
-      "grad_norm": 0.6501268148422241,
-      "learning_rate": 4.055555555555556e-06,
-      "loss": 0.0116,
+      "grad_norm": 1.4347445964813232,
+      "learning_rate": 4.062222222222223e-06,
+      "loss": 0.021,
       "step": 3175
     },
     {
       "epoch": 4.526166902404526,
-      "grad_norm": 0.4661734402179718,
-      "learning_rate": 4.000000000000001e-06,
-      "loss": 0.0125,
+      "grad_norm": 2.03843355178833,
+      "learning_rate": 4.006666666666667e-06,
+      "loss": 0.0278,
       "step": 3200
     },
     {
       "epoch": 4.561527581329561,
-      "grad_norm": 1.581214427947998,
-      "learning_rate": 3.944444444444445e-06,
-      "loss": 0.0126,
+      "grad_norm": 2.6143271923065186,
+      "learning_rate": 3.951111111111112e-06,
+      "loss": 0.0316,
       "step": 3225
     },
     {
       "epoch": 4.596888260254596,
-      "grad_norm": 0.7272769212722778,
-      "learning_rate": 3.88888888888889e-06,
-      "loss": 0.0116,
+      "grad_norm": 2.1741061210632324,
+      "learning_rate": 3.895555555555556e-06,
+      "loss": 0.0277,
       "step": 3250
     },
     {
       "epoch": 4.632248939179632,
-      "grad_norm": 1.0151731967926025,
-      "learning_rate": 3.833333333333334e-06,
-      "loss": 0.0089,
+      "grad_norm": 3.5832366943359375,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 0.025,
       "step": 3275
     },
     {
       "epoch": 4.667609618104668,
-      "grad_norm": 0.4858356714248657,
-      "learning_rate": 3.777777777777778e-06,
-      "loss": 0.0095,
+      "grad_norm": 2.232801675796509,
+      "learning_rate": 3.784444444444445e-06,
+      "loss": 0.0282,
       "step": 3300
     },
     {
       "epoch": 4.702970297029703,
-      "grad_norm": 1.0517410039901733,
-      "learning_rate": 3.7222222222222225e-06,
-      "loss": 0.0107,
+      "grad_norm": 2.736708641052246,
+      "learning_rate": 3.728888888888889e-06,
+      "loss": 0.0308,
       "step": 3325
     },
     {
       "epoch": 4.738330975954739,
-      "grad_norm": 1.0615334510803223,
-      "learning_rate": 3.6666666666666666e-06,
-      "loss": 0.0132,
+      "grad_norm": 2.0116610527038574,
+      "learning_rate": 3.673333333333334e-06,
+      "loss": 0.0305,
       "step": 3350
     },
     {
       "epoch": 4.773691654879774,
-      "grad_norm": 1.1842697858810425,
-      "learning_rate": 3.6111111111111115e-06,
-      "loss": 0.009,
+      "grad_norm": 2.6088578701019287,
+      "learning_rate": 3.617777777777778e-06,
+      "loss": 0.025,
       "step": 3375
     },
     {
       "epoch": 4.809052333804809,
-      "grad_norm": 0.6705470681190491,
-      "learning_rate": 3.555555555555556e-06,
-      "loss": 0.0109,
+      "grad_norm": 1.400682806968689,
+      "learning_rate": 3.5622222222222224e-06,
+      "loss": 0.0248,
       "step": 3400
     },
     {
       "epoch": 4.844413012729844,
-      "grad_norm": 0.22446350753307343,
-      "learning_rate": 3.5e-06,
-      "loss": 0.0113,
+      "grad_norm": 1.956292748451233,
+      "learning_rate": 3.5066666666666673e-06,
+      "loss": 0.0258,
       "step": 3425
     },
     {
       "epoch": 4.879773691654879,
-      "grad_norm": 0.6739686131477356,
-      "learning_rate": 3.444444444444445e-06,
-      "loss": 0.0089,
+      "grad_norm": 2.2373459339141846,
+      "learning_rate": 3.4511111111111113e-06,
+      "loss": 0.0229,
       "step": 3450
     },
     {
       "epoch": 4.915134370579915,
-      "grad_norm": 0.9094479084014893,
-      "learning_rate": 3.3888888888888893e-06,
-      "loss": 0.0096,
+      "grad_norm": 2.9303152561187744,
+      "learning_rate": 3.3955555555555558e-06,
+      "loss": 0.0255,
       "step": 3475
     },
     {
       "epoch": 4.9504950495049505,
-      "grad_norm": 0.486494243144989,
-      "learning_rate": 3.3333333333333333e-06,
-      "loss": 0.0116,
+      "grad_norm": 2.3725924491882324,
+      "learning_rate": 3.3400000000000006e-06,
+      "loss": 0.0289,
       "step": 3500
     },
     {
       "epoch": 4.985855728429986,
-      "grad_norm": 1.4182993173599243,
-      "learning_rate": 3.277777777777778e-06,
-      "loss": 0.011,
+      "grad_norm": 2.325536012649536,
+      "learning_rate": 3.2844444444444447e-06,
+      "loss": 0.0277,
       "step": 3525
     },
     {
       "epoch": 5.021216407355022,
-      "grad_norm": 0.35871076583862305,
-      "learning_rate": 3.2222222222222227e-06,
-      "loss": 0.0087,
+      "grad_norm": 3.026657819747925,
+      "learning_rate": 3.228888888888889e-06,
+      "loss": 0.0184,
       "step": 3550
     },
     {
       "epoch": 5.056577086280057,
-      "grad_norm": 0.08477536588907242,
-      "learning_rate": 3.1666666666666667e-06,
-      "loss": 0.0053,
+      "grad_norm": 1.5165703296661377,
+      "learning_rate": 3.173333333333334e-06,
+      "loss": 0.0148,
       "step": 3575
     },
     {
       "epoch": 5.091937765205092,
-      "grad_norm": 0.2496667057275772,
-      "learning_rate": 3.1111111111111116e-06,
-      "loss": 0.0033,
+      "grad_norm": 1.2629361152648926,
+      "learning_rate": 3.117777777777778e-06,
+      "loss": 0.0104,
       "step": 3600
     },
     {
       "epoch": 5.127298444130127,
-      "grad_norm": 0.21481718122959137,
-      "learning_rate": 3.055555555555556e-06,
-      "loss": 0.0039,
+      "grad_norm": 1.312657117843628,
+      "learning_rate": 3.0622222222222225e-06,
+      "loss": 0.0132,
       "step": 3625
     },
     {
       "epoch": 5.162659123055163,
-      "grad_norm": 1.028740406036377,
-      "learning_rate": 3e-06,
-      "loss": 0.0027,
+      "grad_norm": 1.1759546995162964,
+      "learning_rate": 3.0066666666666674e-06,
+      "loss": 0.0096,
       "step": 3650
     },
     {
       "epoch": 5.198019801980198,
-      "grad_norm": 1.1931138038635254,
-      "learning_rate": 2.944444444444445e-06,
-      "loss": 0.0041,
+      "grad_norm": 0.8515748381614685,
+      "learning_rate": 2.9511111111111114e-06,
+      "loss": 0.0125,
       "step": 3675
     },
     {
       "epoch": 5.233380480905233,
-      "grad_norm": 0.2043699473142624,
-      "learning_rate": 2.888888888888889e-06,
-      "loss": 0.0042,
+      "grad_norm": 1.0795273780822754,
+      "learning_rate": 2.895555555555556e-06,
+      "loss": 0.0134,
       "step": 3700
     },
     {
       "epoch": 5.2687411598302685,
-      "grad_norm": 0.15545345842838287,
-      "learning_rate": 2.8333333333333335e-06,
-      "loss": 0.0036,
+      "grad_norm": 1.2719004154205322,
+      "learning_rate": 2.84e-06,
+      "loss": 0.0104,
       "step": 3725
     },
     {
       "epoch": 5.3041018387553045,
-      "grad_norm": 0.127390518784523,
-      "learning_rate": 2.7777777777777783e-06,
-      "loss": 0.0035,
+      "grad_norm": 0.8354203701019287,
+      "learning_rate": 2.784444444444445e-06,
+      "loss": 0.0106,
       "step": 3750
     },
     {
       "epoch": 5.33946251768034,
-      "grad_norm": 0.2568166255950928,
-      "learning_rate": 2.7222222222222224e-06,
-      "loss": 0.0039,
+      "grad_norm": 2.3398609161376953,
+      "learning_rate": 2.7288888888888893e-06,
+      "loss": 0.0126,
       "step": 3775
     },
     {
       "epoch": 5.374823196605375,
-      "grad_norm": 0.7207896709442139,
-      "learning_rate": 2.666666666666667e-06,
-      "loss": 0.0037,
+      "grad_norm": 2.567042350769043,
+      "learning_rate": 2.6733333333333333e-06,
+      "loss": 0.0109,
       "step": 3800
     },
     {
       "epoch": 5.41018387553041,
-      "grad_norm": 0.2093198150396347,
-      "learning_rate": 2.6111111111111113e-06,
-      "loss": 0.0033,
+      "grad_norm": 1.6744190454483032,
+      "learning_rate": 2.617777777777778e-06,
+      "loss": 0.0117,
       "step": 3825
     },
     {
       "epoch": 5.445544554455446,
-      "grad_norm": 1.7021733522415161,
-      "learning_rate": 2.5555555555555557e-06,
-      "loss": 0.0044,
+      "grad_norm": 1.0200034379959106,
+      "learning_rate": 2.5622222222222226e-06,
+      "loss": 0.0118,
       "step": 3850
     },
     {
       "epoch": 5.480905233380481,
-      "grad_norm": 0.4927394390106201,
-      "learning_rate": 2.5e-06,
-      "loss": 0.0033,
+      "grad_norm": 0.8212645053863525,
+      "learning_rate": 2.5066666666666667e-06,
+      "loss": 0.0106,
       "step": 3875
     },
     {
       "epoch": 5.516265912305516,
-      "grad_norm": 0.16159459948539734,
-      "learning_rate": 2.4444444444444447e-06,
-      "loss": 0.0035,
+      "grad_norm": 0.9394924640655518,
+      "learning_rate": 2.451111111111111e-06,
+      "loss": 0.0103,
       "step": 3900
     },
     {
       "epoch": 5.551626591230551,
-      "grad_norm": 0.25875765085220337,
-      "learning_rate": 2.388888888888889e-06,
-      "loss": 0.0049,
+      "grad_norm": 2.0436089038848877,
+      "learning_rate": 2.3955555555555556e-06,
+      "loss": 0.0177,
       "step": 3925
     },
     {
       "epoch": 5.586987270155587,
-      "grad_norm": 0.22136487066745758,
-      "learning_rate": 2.3333333333333336e-06,
-      "loss": 0.0045,
+      "grad_norm": 1.194536566734314,
+      "learning_rate": 2.3400000000000005e-06,
+      "loss": 0.0105,
       "step": 3950
     },
     {
       "epoch": 5.6223479490806225,
-      "grad_norm": 0.2594265341758728,
-      "learning_rate": 2.277777777777778e-06,
-      "loss": 0.0035,
+      "grad_norm": 1.526948094367981,
+      "learning_rate": 2.2844444444444445e-06,
+      "loss": 0.0097,
       "step": 3975
     },
     {
       "epoch": 5.657708628005658,
-      "grad_norm": 0.46229287981987,
-      "learning_rate": 2.222222222222222e-06,
-      "loss": 0.005,
+      "grad_norm": 1.8187007904052734,
+      "learning_rate": 2.228888888888889e-06,
+      "loss": 0.0109,
       "step": 4000
     },
     {
       "epoch": 5.657708628005658,
-      "eval_loss": 0.2928106486797333,
-      "eval_runtime": 2246.0565,
-      "eval_samples_per_second": 2.396,
-      "eval_steps_per_second": 0.15,
-      "eval_wer": 0.16921149946319933,
+      "eval_loss": 0.30111825466156006,
+      "eval_runtime": 1778.4921,
+      "eval_samples_per_second": 3.026,
+      "eval_steps_per_second": 0.189,
+      "eval_wer": 0.18279056821344786,
       "step": 4000
     },
     {
       "epoch": 5.693069306930693,
-      "grad_norm": 0.6756157279014587,
-      "learning_rate": 2.166666666666667e-06,
-      "loss": 0.0033,
+      "grad_norm": 2.2238926887512207,
+      "learning_rate": 2.1733333333333334e-06,
+      "loss": 0.0096,
       "step": 4025
     },
     {
       "epoch": 5.728429985855728,
-      "grad_norm": 0.2873589098453522,
-      "learning_rate": 2.1111111111111114e-06,
-      "loss": 0.0038,
+      "grad_norm": 1.989168405532837,
+      "learning_rate": 2.117777777777778e-06,
+      "loss": 0.0136,
       "step": 4050
     },
     {
       "epoch": 5.763790664780764,
-      "grad_norm": 0.2609182596206665,
-      "learning_rate": 2.0555555555555555e-06,
-      "loss": 0.0033,
+      "grad_norm": 1.6722816228866577,
+      "learning_rate": 2.0622222222222223e-06,
+      "loss": 0.0107,
       "step": 4075
     },
     {
       "epoch": 5.799151343705799,
-      "grad_norm": 0.47305575013160706,
-      "learning_rate": 2.0000000000000003e-06,
-      "loss": 0.0026,
+      "grad_norm": 1.0947668552398682,
+      "learning_rate": 2.006666666666667e-06,
+      "loss": 0.0083,
       "step": 4100
     },
     {
       "epoch": 5.834512022630834,
-      "grad_norm": 0.11515957117080688,
-      "learning_rate": 1.944444444444445e-06,
-      "loss": 0.0028,
+      "grad_norm": 0.5356242656707764,
+      "learning_rate": 1.9511111111111113e-06,
+      "loss": 0.0108,
       "step": 4125
     },
     {
       "epoch": 5.86987270155587,
-      "grad_norm": 0.48609447479248047,
-      "learning_rate": 1.888888888888889e-06,
-      "loss": 0.0034,
+      "grad_norm": 2.1374714374542236,
+      "learning_rate": 1.8955555555555557e-06,
+      "loss": 0.0107,
       "step": 4150
     },
     {
       "epoch": 5.905233380480905,
-      "grad_norm": 1.1658227443695068,
-      "learning_rate": 1.8333333333333333e-06,
-      "loss": 0.0038,
+      "grad_norm": 0.9695949554443359,
+      "learning_rate": 1.8400000000000002e-06,
+      "loss": 0.0103,
       "step": 4175
     },
     {
       "epoch": 5.9405940594059405,
-      "grad_norm": 0.13836970925331116,
-      "learning_rate": 1.777777777777778e-06,
-      "loss": 0.0044,
+      "grad_norm": 1.3653473854064941,
+      "learning_rate": 1.7844444444444444e-06,
+      "loss": 0.0116,
       "step": 4200
     },
     {
       "epoch": 5.975954738330976,
-      "grad_norm": 0.46859803795814514,
-      "learning_rate": 1.7222222222222224e-06,
-      "loss": 0.0041,
+      "grad_norm": 1.5049793720245361,
+      "learning_rate": 1.728888888888889e-06,
+      "loss": 0.0129,
       "step": 4225
     },
     {
       "epoch": 6.011315417256012,
-      "grad_norm": 0.21296146512031555,
-      "learning_rate": 1.6666666666666667e-06,
-      "loss": 0.0028,
+      "grad_norm": 2.540705442428589,
+      "learning_rate": 1.6733333333333335e-06,
+      "loss": 0.0085,
       "step": 4250
     },
     {
       "epoch": 6.046676096181047,
-      "grad_norm": 0.05787177383899689,
-      "learning_rate": 1.6111111111111113e-06,
-      "loss": 0.002,
+      "grad_norm": 0.5889933705329895,
+      "learning_rate": 1.6177777777777778e-06,
+      "loss": 0.0043,
       "step": 4275
     },
     {
       "epoch": 6.082036775106082,
-      "grad_norm": 0.09653793275356293,
-      "learning_rate": 1.5555555555555558e-06,
-      "loss": 0.002,
+      "grad_norm": 1.6074843406677246,
+      "learning_rate": 1.5622222222222225e-06,
+      "loss": 0.006,
       "step": 4300
     },
     {
       "epoch": 6.117397454031117,
-      "grad_norm": 0.40661904215812683,
-      "learning_rate": 1.5e-06,
-      "loss": 0.0015,
+      "grad_norm": 0.8578146696090698,
+      "learning_rate": 1.506666666666667e-06,
+      "loss": 0.0054,
       "step": 4325
     },
     {
       "epoch": 6.152758132956153,
-      "grad_norm": 0.05121398717164993,
-      "learning_rate": 1.4444444444444445e-06,
-      "loss": 0.0015,
+      "grad_norm": 0.6143340468406677,
+      "learning_rate": 1.4511111111111112e-06,
+      "loss": 0.0043,
       "step": 4350
     },
     {
       "epoch": 6.188118811881188,
-      "grad_norm": 0.06613893806934357,
-      "learning_rate": 1.3888888888888892e-06,
-      "loss": 0.0014,
+      "grad_norm": 0.3723517656326294,
+      "learning_rate": 1.3955555555555556e-06,
+      "loss": 0.0034,
       "step": 4375
     },
     {
       "epoch": 6.223479490806223,
-      "grad_norm": 0.04856628179550171,
-      "learning_rate": 1.3333333333333334e-06,
-      "loss": 0.0023,
+      "grad_norm": 1.0041955709457397,
+      "learning_rate": 1.34e-06,
+      "loss": 0.0059,
       "step": 4400
     },
     {
       "epoch": 6.258840169731259,
-      "grad_norm": 0.11592649668455124,
-      "learning_rate": 1.2777777777777779e-06,
-      "loss": 0.0013,
+      "grad_norm": 1.2664568424224854,
+      "learning_rate": 1.2844444444444445e-06,
+      "loss": 0.0039,
       "step": 4425
     },
     {
       "epoch": 6.294200848656295,
-      "grad_norm": 0.0709163099527359,
-      "learning_rate": 1.2222222222222223e-06,
-      "loss": 0.0017,
+      "grad_norm": 0.259546160697937,
+      "learning_rate": 1.228888888888889e-06,
+      "loss": 0.0051,
       "step": 4450
     },
     {
       "epoch": 6.32956152758133,
-      "grad_norm": 0.05270463228225708,
-      "learning_rate": 1.1666666666666668e-06,
-      "loss": 0.0016,
+      "grad_norm": 0.9940450191497803,
+      "learning_rate": 1.1733333333333335e-06,
+      "loss": 0.0027,
       "step": 4475
     },
     {
       "epoch": 6.364922206506365,
-      "grad_norm": 0.23842215538024902,
-      "learning_rate": 1.111111111111111e-06,
-      "loss": 0.0014,
+      "grad_norm": 0.23329602181911469,
+      "learning_rate": 1.117777777777778e-06,
+      "loss": 0.0038,
       "step": 4500
     },
     {
       "epoch": 6.4002828854314,
-      "grad_norm": 0.046483710408210754,
-      "learning_rate": 1.0555555555555557e-06,
-      "loss": 0.0014,
+      "grad_norm": 0.12971219420433044,
+      "learning_rate": 1.0622222222222222e-06,
+      "loss": 0.0035,
       "step": 4525
     },
     {
       "epoch": 6.435643564356436,
-      "grad_norm": 0.050459932535886765,
-      "learning_rate": 1.0000000000000002e-06,
-      "loss": 0.0021,
+      "grad_norm": 0.4987533688545227,
+      "learning_rate": 1.0066666666666668e-06,
+      "loss": 0.0081,
       "step": 4550
     },
     {
       "epoch": 6.471004243281471,
-      "grad_norm": 0.04769385978579521,
-      "learning_rate": 9.444444444444445e-07,
-      "loss": 0.0014,
+      "grad_norm": 0.27808260917663574,
+      "learning_rate": 9.511111111111111e-07,
+      "loss": 0.0034,
       "step": 4575
     },
     {
       "epoch": 6.506364922206506,
-      "grad_norm": 0.08741884678602219,
-      "learning_rate": 8.88888888888889e-07,
-      "loss": 0.0012,
+      "grad_norm": 0.23258398473262787,
+      "learning_rate": 8.955555555555557e-07,
+      "loss": 0.0033,
       "step": 4600
     },
     {
       "epoch": 6.5417256011315414,
-      "grad_norm": 0.07672213762998581,
-      "learning_rate": 8.333333333333333e-07,
-      "loss": 0.0013,
+      "grad_norm": 0.3250856101512909,
+      "learning_rate": 8.400000000000001e-07,
+      "loss": 0.0038,
       "step": 4625
     },
     {
       "epoch": 6.5770862800565775,
-      "grad_norm": 0.3090445399284363,
-      "learning_rate": 7.777777777777779e-07,
-      "loss": 0.0013,
+      "grad_norm": 0.5452969074249268,
+      "learning_rate": 7.844444444444445e-07,
+      "loss": 0.0033,
       "step": 4650
     },
     {
       "epoch": 6.612446958981613,
-      "grad_norm": 0.06381077319383621,
-      "learning_rate": 7.222222222222222e-07,
-      "loss": 0.0023,
+      "grad_norm": 0.22152584791183472,
+      "learning_rate": 7.28888888888889e-07,
+      "loss": 0.0067,
       "step": 4675
     },
     {
       "epoch": 6.647807637906648,
-      "grad_norm": 0.10808974504470825,
-      "learning_rate": 6.666666666666667e-07,
-      "loss": 0.0017,
+      "grad_norm": 1.5061694383621216,
+      "learning_rate": 6.733333333333334e-07,
+      "loss": 0.0081,
       "step": 4700
     },
     {
       "epoch": 6.683168316831683,
-      "grad_norm": 0.06639378517866135,
-      "learning_rate": 6.111111111111112e-07,
-      "loss": 0.0013,
+      "grad_norm": 0.2032124400138855,
+      "learning_rate": 6.177777777777778e-07,
+      "loss": 0.0042,
       "step": 4725
     },
     {
       "epoch": 6.718528995756719,
-      "grad_norm": 0.05687042698264122,
-      "learning_rate": 5.555555555555555e-07,
-      "loss": 0.0017,
+      "grad_norm": 0.25647231936454773,
+      "learning_rate": 5.622222222222223e-07,
+      "loss": 0.0061,
       "step": 4750
     },
     {
       "epoch": 6.753889674681754,
-      "grad_norm": 0.050000376999378204,
-      "learning_rate": 5.000000000000001e-07,
-      "loss": 0.0012,
+      "grad_norm": 0.27424803376197815,
+      "learning_rate": 5.066666666666667e-07,
+      "loss": 0.0034,
       "step": 4775
     },
     {
       "epoch": 6.789250353606789,
-      "grad_norm": 0.043993644416332245,
-      "learning_rate": 4.444444444444445e-07,
-      "loss": 0.0015,
+      "grad_norm": 0.4183076322078705,
+      "learning_rate": 4.511111111111111e-07,
+      "loss": 0.004,
       "step": 4800
     },
     {
       "epoch": 6.824611032531824,
-      "grad_norm": 0.0865250751376152,
-      "learning_rate": 3.8888888888888895e-07,
-      "loss": 0.0017,
+      "grad_norm": 0.40847113728523254,
+      "learning_rate": 3.9555555555555557e-07,
+      "loss": 0.0029,
       "step": 4825
     },
     {
       "epoch": 6.85997171145686,
-      "grad_norm": 0.06808679550886154,
-      "learning_rate": 3.3333333333333335e-07,
-      "loss": 0.0011,
+      "grad_norm": 0.2560099959373474,
+      "learning_rate": 3.4000000000000003e-07,
+      "loss": 0.0028,
       "step": 4850
     },
     {
       "epoch": 6.8953323903818955,
-      "grad_norm": 0.03820795193314552,
-      "learning_rate": 2.7777777777777776e-07,
-      "loss": 0.0014,
+      "grad_norm": 0.13006103038787842,
+      "learning_rate": 2.844444444444445e-07,
+      "loss": 0.0042,
       "step": 4875
     },
     {
       "epoch": 6.930693069306931,
-      "grad_norm": 0.07484640926122665,
-      "learning_rate": 2.2222222222222224e-07,
-      "loss": 0.0014,
+      "grad_norm": 0.38391485810279846,
+      "learning_rate": 2.2888888888888892e-07,
+      "loss": 0.0029,
       "step": 4900
     },
     {
       "epoch": 6.966053748231966,
-      "grad_norm": 0.07878479361534119,
-      "learning_rate": 1.6666666666666668e-07,
-      "loss": 0.0012,
+      "grad_norm": 0.4023485481739044,
+      "learning_rate": 1.7333333333333335e-07,
+      "loss": 0.0027,
       "step": 4925
     },
     {
       "epoch": 7.001414427157002,
-      "grad_norm": 0.0370086245238781,
-      "learning_rate": 1.1111111111111112e-07,
-      "loss": 0.0017,
+      "grad_norm": 0.1441173106431961,
+      "learning_rate": 1.1777777777777778e-07,
+      "loss": 0.0042,
       "step": 4950
     },
     {
       "epoch": 7.036775106082037,
-      "grad_norm": 0.04006652534008026,
-      "learning_rate": 5.555555555555556e-08,
-      "loss": 0.0012,
+      "grad_norm": 0.18802694976329803,
+      "learning_rate": 6.222222222222223e-08,
+      "loss": 0.0033,
       "step": 4975
     },
     {
       "epoch": 7.072135785007072,
-      "grad_norm": 0.04126604646444321,
-      "learning_rate": 0.0,
-      "loss": 0.0009,
+      "grad_norm": 0.14834970235824585,
+      "learning_rate": 6.666666666666667e-09,
+      "loss": 0.0016,
       "step": 5000
     },
     {
       "epoch": 7.072135785007072,
-      "eval_loss": 0.3181316554546356,
-      "eval_runtime": 2177.6024,
-      "eval_samples_per_second": 2.471,
-      "eval_steps_per_second": 0.155,
-      "eval_wer": 0.16825718716449958,
+      "eval_loss": 0.3134021461009979,
+      "eval_runtime": 1756.5453,
+      "eval_samples_per_second": 3.063,
+      "eval_steps_per_second": 0.192,
+      "eval_wer": 0.17463915066205415,
       "step": 5000
     },
     {
       "epoch": 7.072135785007072,
       "step": 5000,
-      "total_flos": 5.4329055670370304e+20,
-      "train_loss": 0.09240171913430095,
-      "train_runtime": 60495.4593,
-      "train_samples_per_second": 2.645,
-      "train_steps_per_second": 0.083
+      "total_flos": 2.7263703386750976e+20,
+      "train_loss": 0.12853609090298415,
+      "train_runtime": 49407.2097,
+      "train_samples_per_second": 3.238,
+      "train_steps_per_second": 0.101
     }
   ],
   "logging_steps": 25,
@@ -1480,7 +1480,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 5.4329055670370304e+20,
+  "total_flos": 2.7263703386750976e+20,
   "train_batch_size": 16,
   "trial_name": null,
   "trial_params": null