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: 0.9806
+- Loss: 0.6364
 
 ## Model description
 
@@ -40,24 +40,23 @@ More information needed
 
 The following hyperparameters were used during training:
 - learning_rate: 0.001
-- train_batch_size: 8
-- eval_batch_size: 8
+- train_batch_size: 16
+- eval_batch_size: 24
 - seed: 42
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
-- lr_scheduler_warmup_steps: 50
-- num_epochs: 1
+- lr_scheduler_warmup_steps: 250
+- num_epochs: 3
 - mixed_precision_training: Native AMP
 
 ### Training results
 
 | Training Loss | Epoch | Step | Validation Loss |
 |:-------------:|:-----:|:----:|:---------------:|
-| 1.1346        | 0.2   | 100  | 1.2898          |
-| 1.1269        | 0.4   | 200  | 1.1432          |
-| 1.0896        | 0.6   | 300  | 1.0504          |
-| 1.0369        | 0.8   | 400  | 1.0032          |
-| 1.0069        | 1.0   | 500  | 0.9806          |
+| 0.9605        | 0.65  | 1000 | 0.9080          |
+| 0.7138        | 1.29  | 2000 | 0.7496          |
+| 0.6426        | 1.94  | 3000 | 0.6741          |
+| 0.5076        | 2.58  | 4000 | 0.6364          |
 
 
 ### Framework versions

runs/Mar12_03-55-13_88b779dc21b0/events.out.tfevents.1710215718.88b779dc21b0.26.0

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9ab33a0c1aa8acb7108afaa1b2382b62de9a906cdbd99daa540eedc6285f3fc3
-size 15439
+oid sha256:e9cf0b5a713afb77ac1012f53810a82b86ee65967643919215e1a8f5ab60a1b4
+size 17059