---

language: ja

license: cc-by-sa-4.0

datasets:

- YACIS corpus

---

# yacis-electra-small

This is [ELECTRA](https://github.com/google-research/electra) Small model for Japanese pretrained on 354 million sentences / 5.6 billion words of [YACIS](https://github.com/ptaszynski/yacis-corpus) blog corpus.

The corpus was tokenized for pretraining with [MeCab](https://taku910.github.io/mecab/). Subword tokenization was peroformed with WordPiece. 

## Model architecture

This model uses the original ELECTRA Small model; 12 layers, 128 dimensions of hidden states, and 12 attention heads.

Vocabulary size was 32,000 tokens.

## Licenses

The pretrained model with all attached files is distributed under the terms of the [CC BY-SA 4.0](https://creativecommons.org/licenses/by-sa/4.0/deed.en) license. 
<a rel="license" href="http://creativecommons.org/licenses/by-sa/4.0/"><img alt="Creative Commons License" style="border-width:0" src="https://i.creativecommons.org/l/by-sa/4.0/88x31.png" /></a><br />This work is licensed under a <a rel="license" href="http://creativecommons.org/licenses/by-sa/4.0/">Creative Commons Attribution-ShareAlike 4.0 International License</a>.