File size: 1,279 Bytes
a7caa81 f4c2042 a7caa81 7447ea4 a7caa81 46205ff a7caa81 fa619db |
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 28 29 30 31 32 33 34 35 36 |
---
language:
- "ain"
tags:
- "ainu"
- "token-classification"
- "pos"
- "dependency-parsing"
base_model: KoichiYasuoka/roberta-base-ainu-upos
license: "cc-by-sa-4.0"
pipeline_tag: "token-classification"
widget:
- text: "itak=as awa pon rupne aynu ene itaki"
- text: "イタカㇱ アワ ポン ルㇷ゚ネ アイヌ エネ イタキ"
- text: "итакас ава пон рубне айну эне итакі"
---
# roberta-base-ainu-ud-goeswith
## Model Description
This is a RoBERTa model pre-trained on Ainu texts (in カタカナ, Roman, and Кириллица) for POS-tagging and dependency-parsing (using `goeswith` for subwords), derived from [roberta-base-ainu-upos](https://huggingface.co/KoichiYasuoka/roberta-base-ainu-upos).
## How to Use
```py
from transformers import pipeline
nlp=pipeline("universal-dependencies","KoichiYasuoka/roberta-base-ainu-ud-goeswith",trust_remote_code=True,aggregation_strategy="simple")
print(nlp("itak=as awa pon rupne aynu ene itaki"))
```
## Reference
安岡孝一: [ローマ字・カタカナ・キリル文字併用アイヌ語RoBERTa・DeBERTaモデルの開発](http://id.nii.ac.jp/1001/00224072/), 情報処理学会研究報告, Vol.2023-CH-131『人文科学とコンピュータ』, No.7 (2023年2月18日), pp.1-7.
|