Bauwens's picture
Update README.md
bf1af7b verified
|
raw
history blame contribute delete
No virus
314 Bytes
---
license: apache-2.0
---
To load the tokeniser for this model, do *not* use `AutoTokenizer`. Instead, install [TkTkT](https://github.com/bauwenst/TkTkT) and run:
```python
from tktkt.models.bpe.knockout import BPEKnockout
tokenizer = BPEKnockout.from_pretrained("Bauwens/RoBERTa-nl_BPE_30k_BPE-knockout_9k")
```