Update README.md
Browse files
README.md
CHANGED
|
@@ -44,5 +44,8 @@ Fine-tuned [facebook/wav2vec2-base-fr-voxpopuli-v2](https://huggingface.co/faceb
|
|
| 44 |
- Second warmup boost to 3e-5 for 3% of total updates
|
| 45 |
- Constant at 3e-5 for 12% of total updates
|
| 46 |
- Linear decrease to 1e-7 for remaining 14% of updates
|
|
|
|
|
|
|
|
|
|
|
|
|
| 47 |
|
| 48 |
-
When using this model, make sure that your speech input is sampled at 16kHz.
|
|
|
|
| 44 |
- Second warmup boost to 3e-5 for 3% of total updates
|
| 45 |
- Constant at 3e-5 for 12% of total updates
|
| 46 |
- Linear decrease to 1e-7 for remaining 14% of updates
|
| 47 |
+
|
| 48 |
+
- The set of hyperparameters used for training are those found in [wav2vec2 paper](https://arxiv.org/pdf/2006.11477.pdf), Table 6.
|
| 49 |
+
|
| 50 |
+
When using this model, make sure that your speech input is **sampled at 16kHz**.
|
| 51 |
|
|
|