End of training

README.md

@@ -20,7 +20,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on the Common Voice 16.1 dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.8943
+- Loss: 0.9023
 
 ## Model description
 
@@ -48,7 +48,7 @@ The following hyperparameters were used during training:
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
 - lr_scheduler_warmup_ratio: 0.1
-- num_epochs: 4
+- num_epochs: 5
 - mixed_precision_training: Native AMP
 
 ### Training results
@@ -57,7 +57,7 @@ The following hyperparameters were used during training:
 |:-------------:|:-----:|:-----:|:---------------:|
 | 0.9794        | 1.0   | 2500  | 1.0233          |
 | 0.7059        | 3.0   | 7500  | 0.8989          |
-| 0.5392        | 4.0   | 10000 | 0.8943          |
+| 0.5047        | 5.0   | 12500 | 0.9023          |
 
 
 ### Framework versions

runs/Aug11_19-03-03_dd311876642e/events.out.tfevents.1723402985.dd311876642e.25.0

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:68ff41bd527c85f933d655172036d9ef9aa3f3ccbb25c9811190664fd4aca56d
-size 26544
+oid sha256:3a183bc3d236bf32d0a95353cb81242a8f9b8c1627261c7a3ddbf8802082fa96
+size 27169