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: 0.8856
+- Loss: 0.9073
 
 ## 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: 5
+- num_epochs: 7
 - mixed_precision_training: Native AMP
 
 ### Training results
@@ -59,7 +59,9 @@ The following hyperparameters were used during training:
 | 0.846         | 2.0   | 5056  | 0.9359          |
 | 0.7486        | 3.0   | 7585  | 0.8879          |
 | 0.6056        | 4.0   | 10112 | 0.8839          |
-| 0.4393        | 5.0   | 12640 | 0.8856          |
+| 0.4569        | 5.0   | 12640 | 0.9000          |
+| 0.4929        | 6.0   | 15169 | 0.8907          |
+| 0.3637        | 7.0   | 17696 | 0.9073          |
 
 
 ### Framework versions

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:216c52a2fe825cd41255cae624f7744922421667c1d94ec2657f0d4eeaff236f
+oid sha256:a070e64524ee81871c0c340015de5ab5c3f449cd00eff279d3e61ad5ef75a356
 size 19548472

runs/Sep02_02-53-56_593a571355c3/events.out.tfevents.1725245638.593a571355c3.24.0

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5fbeaa97a576ed3d8dc9e0d7ad2d48655bd57c3314dfe3e59ef8d20e54da1de0
-size 48366
+oid sha256:79a358876eb6e5b75eca450b30233b3b83f9ae3e190e93acf24ecf6c5c36664d
+size 50507