Model save

README.md

@@ -1,4 +1,6 @@
 ---
+license: apache-2.0
+base_model: openai/whisper-tiny
 tags:
 - generated_from_trainer
 metrics:
@@ -13,10 +15,10 @@ should probably proofread and complete it, then remove this comment. -->
 
 # whisper-tiny-hi2
 
-This model was trained from scratch on the None dataset.
+This model is a fine-tuned version of [openai/whisper-tiny](https://huggingface.co/openai/whisper-tiny) on the None dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.4554
-- Wer: 56.1182
+- Loss: 0.3163
+- Wer: 41.5276
 
 ## Model description
 
@@ -35,21 +37,24 @@ More information needed
 ### Training hyperparameters
 
 The following hyperparameters were used during training:
-- learning_rate: 7.8e-08
+- learning_rate: 1e-05
 - train_batch_size: 32
 - eval_batch_size: 1
 - seed: 42
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
-- training_steps: 500
+- lr_scheduler_warmup_steps: 500
+- training_steps: 2000
 - mixed_precision_training: Native AMP
 
 ### Training results
 
 | Training Loss | Epoch | Step | Validation Loss | Wer     |
 |:-------------:|:-----:|:----:|:---------------:|:-------:|
-| 0.4285        | 1.0   | 250  | 0.4558          | 56.3045 |
-| 0.4273        | 2.0   | 500  | 0.4554          | 56.1182 |
+| 1.0738        | 2.0   | 500  | 0.4355          | 53.2052 |
+| 0.3349        | 4.0   | 1000 | 0.3382          | 44.7117 |
+| 0.242         | 6.0   | 1500 | 0.3177          | 42.8741 |
+| 0.2009        | 8.0   | 2000 | 0.3163          | 41.5276 |
 
 
 ### Framework versions