End of training

README.md

@@ -6,19 +6,19 @@ tags:
 metrics:
 - wer
 model-index:
-- name: Whisper-opus
+- name: opus-base
   results: []
 ---
 
 <!-- This model card has been generated automatically according to the information the Trainer had access to. You
 should probably proofread and complete it, then remove this comment. -->
 
-# Whisper-opus
+# opus-base
 
 This model is a fine-tuned version of [openai/whisper-base](https://huggingface.co/openai/whisper-base) on an unknown dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.0401
-- Wer: 26.8161
+- Loss: 0.0250
+- Wer: 18.2886
 
 ## Model description
 
@@ -38,23 +38,21 @@ More information needed
 
 The following hyperparameters were used during training:
 - learning_rate: 1e-05
-- train_batch_size: 16
+- train_batch_size: 12
 - eval_batch_size: 2
 - seed: 42
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
-- lr_scheduler_warmup_steps: 600
-- training_steps: 4000
+- lr_scheduler_warmup_steps: 500
+- training_steps: 2000
 - mixed_precision_training: Native AMP
 
 ### Training results
 
 | Training Loss | Epoch  | Step | Validation Loss | Wer     |
 |:-------------:|:------:|:----:|:---------------:|:-------:|
-| 0.1123        | 0.4912 | 1000 | 0.0716          | 43.9402 |
-| 0.052         | 0.9823 | 2000 | 0.0488          | 31.7834 |
-| 0.0402        | 1.4735 | 3000 | 0.0425          | 27.7498 |
-| 0.0367        | 1.9646 | 4000 | 0.0401          | 26.8161 |
+| 0.0293        | 0.3668 | 1000 | 0.0277          | 22.5839 |
+| 0.0283        | 0.7337 | 2000 | 0.0250          | 18.2886 |
 
 
 ### Framework versions