This is ELECTRA Small model for Japanese pretrained on 354 million sentences / 5.6 billion words of YACIS blog corpus.
The corpus was tokenized with MeCab.
