Update Luong-Word-Model/README.md
Browse files
Luong-Word-Model/README.md
CHANGED
|
@@ -7,7 +7,7 @@ CRNN with Luong multiplicative attention between BiLSTM layers 2 and 3.
|
|
| 7 |
- CNN backbone: 6 convolutional blocks (max 256 channels)
|
| 8 |
- Recurrent block: 3 BiLSTM layers, hidden = 160 per direction
|
| 9 |
- Decoding: Connectionist Temporal Classification (CTC)
|
| 10 |
-
- Parameters: 3,
|
| 11 |
|
| 12 |
## Test Performance (DASTNUS, seed 42)
|
| 13 |
| Metric | Value |
|
|
|
|
| 7 |
- CNN backbone: 6 convolutional blocks (max 256 channels)
|
| 8 |
- Recurrent block: 3 BiLSTM layers, hidden = 160 per direction
|
| 9 |
- Decoding: Connectionist Temporal Classification (CTC)
|
| 10 |
+
- Parameters: 3,915,345
|
| 11 |
|
| 12 |
## Test Performance (DASTNUS, seed 42)
|
| 13 |
| Metric | Value |
|