Create README.md
#1
by
lenguist
- opened
README.md
ADDED
|
@@ -0,0 +1,19 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
language:
|
| 3 |
+
- uk
|
| 4 |
+
library_name: transformers
|
| 5 |
+
pipeline_tag: text2text-generation
|
| 6 |
+
tags:
|
| 7 |
+
- grammatical error correction
|
| 8 |
+
- GEC
|
| 9 |
+
- Grammar
|
| 10 |
+
---
|
| 11 |
+
This model was trained by Pravopysnyk team for the Ukrainian NLP shared task in Ukrainian grammar correction. The model is MBart-50-large
|
| 12 |
+
set to ukr-to-ukr translation task finetuned on UA-GEC augmented by custom dataset generated using our synthetic error generation.
|
| 13 |
+
The code for error generation will be uploaded on github soon, and the detailed procedure is described in our paper. For this model we added to ua-gec:-
|
| 14 |
+
5k sentences generated by round translaion (ukr-rus-ukr)
|
| 15 |
+
10k sentences using our punctuation error generation script
|
| 16 |
+
2k of dilution (just fully correct sentences sampled from our dataset)
|
| 17 |
+
10k of russism (errors generated using our russism error generation).
|
| 18 |
+
|
| 19 |
+
Hope you find this description helpful!
|