Commit
•
3fc01ce
1
Parent(s):
28df4e5
Update README
Browse files
README.md
CHANGED
|
@@ -34,6 +34,8 @@ You can fine-tune this model on downstream tasks.
|
|
| 34 |
|
| 35 |
The input text should be segmented into words by [Juman++](https://github.com/ku-nlp/jumanpp) in advance. Juman++ 2.0.0-rc3 was used for pretraining. Each word is tokenized into tokens by [sentencepiece](https://github.com/google/sentencepiece).
|
| 36 |
|
|
|
|
|
|
|
| 37 |
## Vocabulary
|
| 38 |
|
| 39 |
The vocabulary consists of 32000 tokens including words ([JumanDIC](https://github.com/ku-nlp/JumanDIC)) and subwords induced by the unigram language model of [sentencepiece](https://github.com/google/sentencepiece).
|
|
|
|
| 34 |
|
| 35 |
The input text should be segmented into words by [Juman++](https://github.com/ku-nlp/jumanpp) in advance. Juman++ 2.0.0-rc3 was used for pretraining. Each word is tokenized into tokens by [sentencepiece](https://github.com/google/sentencepiece).
|
| 36 |
|
| 37 |
+
`BertJapaneseTokenizer` now supports automatic `JumanppTokenizer` and `SentencepieceTokenizer`. You can use [this model](https://huggingface.co/nlp-waseda/roberta-large-japanese-with-auto-jumanpp) without any data preprocessing.
|
| 38 |
+
|
| 39 |
## Vocabulary
|
| 40 |
|
| 41 |
The vocabulary consists of 32000 tokens including words ([JumanDIC](https://github.com/ku-nlp/JumanDIC)) and subwords induced by the unigram language model of [sentencepiece](https://github.com/google/sentencepiece).
|