luke-japanese-large / README.md
ikuyamada's picture
update README
f650a63
---
language: ja
thumbnail: https://github.com/studio-ousia/luke/raw/master/resources/luke_logo.png
tags:
- luke
- named entity recognition
- entity typing
- relation classification
- question answering
license: apache-2.0
---
## luke-japanese-large
**luke-japanese** is the Japanese version of **LUKE** (**L**anguage
**U**nderstanding with **K**nowledge-based **E**mbeddings), a pre-trained
_knowledge-enhanced_ contextualized representation of words and entities. LUKE
treats words and entities in a given text as independent tokens, and outputs
contextualized representations of them. Please refer to our
[GitHub repository](https://github.com/studio-ousia/luke) for more details and
updates.
This model contains Wikipedia entity embeddings which are not used in general
NLP tasks. Please use the
[lite version](https://huggingface.co/studio-ousia/luke-japanese-large-lite/)
for tasks that do not use Wikipedia entities as inputs.
**luke-japanese**は、単語とエンティティの知識拡張型訓練済み Transformer モデル**LUKE**の日本語版です。LUKE は単語とエンティティを独立したトークンとして扱い、これらの文脈を考慮した表現を出力します。詳細については、[GitHub リポジトリ](https://github.com/studio-ousia/luke)を参照してください。
このモデルは、通常の NLP タスクでは使われない Wikipedia エンティティのエンベディングを含んでいます。単語の入力のみを使うタスクには、[lite version](https://huggingface.co/studio-ousia/luke-japanese-large-lite/)を使用してください。
### Experimental results on JGLUE
The experimental results evaluated on the dev set of
[JGLUE](https://github.com/yahoojapan/JGLUE) is shown as follows:
| Model | MARC-ja | JSTS | JNLI | JCommonsenseQA |
| ----------------------------- | --------- | ------------------- | --------- | -------------- |
| | acc | Pearson/Spearman | acc | acc |
| **LUKE Japanese large** | **0.965** | **0.932**/**0.902** | **0.927** | 0.893 |
| _Baselines:_ | |
| Tohoku BERT large | 0.955 | 0.913/0.872 | 0.900 | 0.816 |
| Waseda RoBERTa large (seq128) | 0.954 | 0.930/0.896 | 0.924 | **0.907** |
| Waseda RoBERTa large (seq512) | 0.961 | 0.926/0.892 | 0.926 | 0.891 |
| XLM RoBERTa large | 0.964 | 0.918/0.884 | 0.919 | 0.840 |
The baseline scores are obtained from
[here](https://github.com/yahoojapan/JGLUE/blob/a6832af23895d6faec8ecf39ec925f1a91601d62/README.md).
### Citation
```latex
@inproceedings{yamada2020luke,
title={LUKE: Deep Contextualized Entity Representations with Entity-aware Self-attention},
author={Ikuya Yamada and Akari Asai and Hiroyuki Shindo and Hideaki Takeda and Yuji Matsumoto},
booktitle={EMNLP},
year={2020}
}
```