update model card README.md
Browse files
README.md
CHANGED
|
@@ -12,15 +12,10 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 12 |
|
| 13 |
# wav2vec2-base-timit-demo-colab
|
| 14 |
|
| 15 |
-
This model is a fine-tuned version of [facebook/
|
| 16 |
It achieves the following results on the evaluation set:
|
| 17 |
-
-
|
| 18 |
-
-
|
| 19 |
-
- eval_runtime: 30.9692
|
| 20 |
-
- eval_samples_per_second: 54.248
|
| 21 |
-
- eval_steps_per_second: 6.781
|
| 22 |
-
- epoch: 4.0
|
| 23 |
-
- step: 500
|
| 24 |
|
| 25 |
## Model description
|
| 26 |
|
|
@@ -49,6 +44,33 @@ The following hyperparameters were used during training:
|
|
| 49 |
- num_epochs: 30
|
| 50 |
- mixed_precision_training: Native AMP
|
| 51 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 52 |
### Framework versions
|
| 53 |
|
| 54 |
- Transformers 4.15.0
|
|
|
|
| 12 |
|
| 13 |
# wav2vec2-base-timit-demo-colab
|
| 14 |
|
| 15 |
+
This model is a fine-tuned version of [facebook/hubert-large-ls960-ft](https://huggingface.co/facebook/hubert-large-ls960-ft) on the None dataset.
|
| 16 |
It achieves the following results on the evaluation set:
|
| 17 |
+
- Loss: 0.0280
|
| 18 |
+
- Wer: 0.0082
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 19 |
|
| 20 |
## Model description
|
| 21 |
|
|
|
|
| 44 |
- num_epochs: 30
|
| 45 |
- mixed_precision_training: Native AMP
|
| 46 |
|
| 47 |
+
### Training results
|
| 48 |
+
|
| 49 |
+
| Training Loss | Epoch | Step | Validation Loss | Wer |
|
| 50 |
+
|:-------------:|:-----:|:-----:|:---------------:|:------:|
|
| 51 |
+
| 0.1152 | 1.42 | 500 | 0.0416 | 0.0159 |
|
| 52 |
+
| 0.0803 | 2.83 | 1000 | 0.0372 | 0.0144 |
|
| 53 |
+
| 0.0672 | 4.25 | 1500 | 0.0345 | 0.0119 |
|
| 54 |
+
| 0.0564 | 5.67 | 2000 | 0.0338 | 0.0106 |
|
| 55 |
+
| 0.0513 | 7.08 | 2500 | 0.0307 | 0.0100 |
|
| 56 |
+
| 0.0448 | 8.5 | 3000 | 0.0343 | 0.0098 |
|
| 57 |
+
| 0.0374 | 9.92 | 3500 | 0.0300 | 0.0084 |
|
| 58 |
+
| 0.0368 | 11.33 | 4000 | 0.0314 | 0.0086 |
|
| 59 |
+
| 0.0388 | 12.75 | 4500 | 0.0283 | 0.0089 |
|
| 60 |
+
| 0.0277 | 14.16 | 5000 | 0.0302 | 0.0089 |
|
| 61 |
+
| 0.0298 | 15.58 | 5500 | 0.0298 | 0.0089 |
|
| 62 |
+
| 0.0271 | 17.0 | 6000 | 0.0320 | 0.0098 |
|
| 63 |
+
| 0.024 | 18.41 | 6500 | 0.0286 | 0.0088 |
|
| 64 |
+
| 0.0236 | 19.83 | 7000 | 0.0284 | 0.0084 |
|
| 65 |
+
| 0.0238 | 21.25 | 7500 | 0.0290 | 0.0086 |
|
| 66 |
+
| 0.0227 | 22.66 | 8000 | 0.0284 | 0.0093 |
|
| 67 |
+
| 0.0198 | 24.08 | 8500 | 0.0280 | 0.0088 |
|
| 68 |
+
| 0.0225 | 25.5 | 9000 | 0.0281 | 0.0086 |
|
| 69 |
+
| 0.018 | 26.91 | 9500 | 0.0280 | 0.0082 |
|
| 70 |
+
| 0.0178 | 28.33 | 10000 | 0.0280 | 0.0082 |
|
| 71 |
+
| 0.0209 | 29.75 | 10500 | 0.0280 | 0.0082 |
|
| 72 |
+
|
| 73 |
+
|
| 74 |
### Framework versions
|
| 75 |
|
| 76 |
- Transformers 4.15.0
|