File size: 506 Bytes
66e7f4a 039ea66 66e7f4a ef45ae6 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 |
---
language: ko
tags:
- bart
license: mit
---
## KoBART-base-v2
With the addition of chatting data, the model is trained to handle the semantics of sequences longer than KoBART.
```python
from transformers import PreTrainedTokenizerFast, BartModel
tokenizer = PreTrainedTokenizerFast.from_pretrained('hyunwoongko/kobart')
model = BartModel.from_pretrained('hyunwoongko/kobart')
```
### Performance
NSMC
- acc. : 0.901
### hyunwoongko/kobart
- Added bos/eos post processor
- Removed token_type_ids
|