End of training

README.md

@@ -7,22 +7,9 @@ tags:
 - generated_from_trainer
 datasets:
 - formospeech/tat_asr_aligned
-metrics:
-- wer
 model-index:
 - name: Whisper Tiny Taiwanese Android
-  results:
-  - task:
-      name: Automatic Speech Recognition
-      type: automatic-speech-recognition
-    dataset:
-      name: TAT ASR Aligned
-      type: formospeech/tat_asr_aligned
-      args: 'config: taiwanese, split: test'
-    metrics:
-    - name: Wer
-      type: wer
-      value: 46.86953383124636
+  results: []
 ---
 
 <!-- This model card has been generated automatically according to the information the Trainer had access to. You
@@ -32,8 +19,8 @@ should probably proofread and complete it, then remove this comment. -->
 
 This model is a fine-tuned version of [openai/whisper-tiny](https://huggingface.co/openai/whisper-tiny) on the TAT ASR Aligned dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.9412
-- Wer: 46.8695
+- Loss: 0.5581
+- Cer: 10.2914
 
 ## Model description
 
@@ -52,40 +39,30 @@ More information needed
 ### Training hyperparameters
 
 The following hyperparameters were used during training:
-- learning_rate: 0.001
+- learning_rate: 0.0001
 - train_batch_size: 64
 - eval_batch_size: 32
 - seed: 42
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
-- lr_scheduler_warmup_steps: 1362
-- training_steps: 13620
+- lr_scheduler_warmup_steps: 681
+- training_steps: 6810
 - mixed_precision_training: Native AMP
 
 ### Training results
 
-| Training Loss | Epoch   | Step  | Validation Loss | Wer     |
-|:-------------:|:-------:|:-----:|:---------------:|:-------:|
-| 0.4657        | 0.9985  | 681   | 0.5472          | 63.0785 |
-| 0.4955        | 1.9971  | 1362  | 0.6142          | 66.7046 |
-| 0.3604        | 2.9956  | 2043  | 0.5341          | 59.9550 |
-| 0.2486        | 3.9941  | 2724  | 0.4831          | 56.0985 |
-| 0.1743        | 4.9927  | 3405  | 0.4750          | 54.1050 |
-| 0.1358        | 5.9912  | 4086  | 0.4915          | 51.4590 |
-| 0.1023        | 6.9897  | 4767  | 0.5325          | 52.2309 |
-| 0.0769        | 7.9883  | 5448  | 0.5405          | 50.7483 |
-| 0.0636        | 8.9868  | 6129  | 0.5518          | 50.9815 |
-| 0.0473        | 9.9853  | 6810  | 0.5817          | 49.4905 |
-| 0.0373        | 10.9839 | 7491  | 0.5983          | 49.9986 |
-| 0.0275        | 11.9824 | 8172  | 0.6474          | 50.0847 |
-| 0.0206        | 12.9809 | 8853  | 0.7030          | 50.2652 |
-| 0.0156        | 13.9795 | 9534  | 0.7327          | 49.8459 |
-| 0.009         | 14.9780 | 10215 | 0.7506          | 48.4244 |
-| 0.0071        | 15.9765 | 10896 | 0.7994          | 48.5271 |
-| 0.003         | 16.9751 | 11577 | 0.8406          | 47.6747 |
-| 0.0014        | 17.9736 | 12258 | 0.8839          | 47.4665 |
-| 0.0003        | 18.9721 | 12939 | 0.9245          | 46.7890 |
-| 0.0001        | 19.9707 | 13620 | 0.9412          | 46.8695 |
+| Training Loss | Epoch  | Step | Validation Loss | Cer     |
+|:-------------:|:------:|:----:|:---------------:|:-------:|
+| 0.3523        | 0.9985 | 681  | 0.4273          | 15.0771 |
+| 0.2104        | 1.9971 | 1362 | 0.3778          | 12.2058 |
+| 0.1224        | 2.9956 | 2043 | 0.3942          | 11.9977 |
+| 0.0738        | 3.9941 | 2724 | 0.4164          | 11.9422 |
+| 0.0423        | 4.9927 | 3405 | 0.4579          | 11.4839 |
+| 0.0216        | 5.9912 | 4086 | 0.4818          | 11.3165 |
+| 0.0107        | 6.9897 | 4767 | 0.5189          | 10.9872 |
+| 0.0036        | 7.9883 | 5448 | 0.5398          | 10.8198 |
+| 0.0012        | 8.9868 | 6129 | 0.5509          | 10.3179 |
+| 0.0006        | 9.9853 | 6810 | 0.5581          | 10.2914 |
 
 
 ### Framework versions