End of training

README.md

@@ -24,7 +24,7 @@ model-index:
     metrics:
     - name: Wer
       type: wer
-      value: 25.31486146095718
+      value: 25.56675062972292
 ---
 
 <!-- This model card has been generated automatically according to the information the Trainer had access to. You
@@ -34,8 +34,8 @@ 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 parlaSmall_subset dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.4923
-- Wer: 25.3149
+- Loss: 0.5479
+- Wer: 25.5668
 
 ## Model description
 
@@ -55,22 +55,24 @@ More information needed
 
 The following hyperparameters were used during training:
 - learning_rate: 1e-05
-- train_batch_size: 2
+- train_batch_size: 1
 - eval_batch_size: 8
 - seed: 42
-- gradient_accumulation_steps: 8
+- gradient_accumulation_steps: 16
 - total_train_batch_size: 16
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
 - lr_scheduler_warmup_steps: 500
-- training_steps: 1000
+- training_steps: 3000
 - mixed_precision_training: Native AMP
 
 ### Training results
 
 | Training Loss | Epoch | Step | Validation Loss | Wer     |
 |:-------------:|:-----:|:----:|:---------------:|:-------:|
-| 0.0004        | 32.52 | 1000 | 0.4923          | 25.3149 |
+| 0.0003        | 32.52 | 1000 | 0.5035          | 26.3224 |
+| 0.0001        | 65.04 | 2000 | 0.5380          | 25.8186 |
+| 0.0001        | 97.56 | 3000 | 0.5479          | 25.5668 |
 
 
 ### Framework versions