Commit
·
a3923ab
1
Parent(s):
c56158d
Update README.md
Browse files
README.md
CHANGED
|
@@ -34,4 +34,5 @@ The model works by:
|
|
| 34 |
It was trained on a parallel corpus (corrupted + fixed sentence) with CTC loss.
|
| 35 |
On our test dataset, it reduces OCR errors by 41%.
|
| 36 |
|
|
|
|
| 37 |
Training details: in [this post](https://habr.com/ru/articles/744972/) (in Russian).
|
|
|
|
| 34 |
It was trained on a parallel corpus (corrupted + fixed sentence) with CTC loss.
|
| 35 |
On our test dataset, it reduces OCR errors by 41%.
|
| 36 |
|
| 37 |
+
Training code: [here](https://github.com/slone-nlp/bashkort-spellcheker/blob/master/experiments/06_ctc_bert.ipynb).
|
| 38 |
Training details: in [this post](https://habr.com/ru/articles/744972/) (in Russian).
|