
A list of pre-trained BERT models for Japanese with word/subword tokenization + vocabulary construction algorithm information
Description in other languages
Statistics
Statistics as of Nov 17, 2026
GitHub stars
132
±0 (Oct 13〜)
Last commit
Mar 2023
Created on GitHub
Jul 26, 2020
Listed
—
Listed since the page was created
Stars over time
Show as a table
| Month | Stars |
|---|---|
| Dec 2025 | 131 |
| Jan 2026 | 131 |
| Feb 2026 | 131 |
| Mar 2026 | 131 |
| Apr 2026 | 132 |
| May 2026 | 132 |
| Jun 2026 | 132 |
| Jul 2026 | 132 |
| Aug 2026 | 132 |
| Sep 2026 | 132 |
| Oct 2026 | 132 |
| Nov 2026 | 132 |
Last snapshot of each month. Star counts of 1,000 or more are rounded to the nearest 100.
Popular in the same category
awesome-japanese-llmllm-jp1.4kSummary of Japanese LLM (Open Source)
tuning_playbook_jaValkyrja3607192A playbook for systematically maximizing the performance of deep learning models.
japanese-pitch-accent-resourcesolety129Trying to consolidate japanese phonetic, and in particular pitch accent resources into one list
dataset-listikegami-yukino120lists of text corpus and more (mainly Japanese)
GEC-Info-jagotutiyan14Repository for collecting and categorizing Japanese literature on correcting grammar errors.
Resources in similar research fields
bert-bookstockmarkteam265"Introduction to Natural Language Processing with BERT: Practical Programming using Transformers" Support Page.Shared: Language Models・Semantic Text Processing
BERT_Japanese_Google_ColaboratoryYutaroOgawa29Google Colaboratoryで日本語のBERTを動かす方法です。Shared: Language Models・Semantic Text Processing- modernbert-ja-310msbintuitions36kModernBERT‑Ja‑310M is a Japanese‑language BERT variant that blends local‑global attention with RoPE, trained on 4.09 T tokens of Japanese/English text, supports a 102 400‑word vocabulary, 8 192‑token sequences, and is optimized for Flash Attention 2.HF modelShared: Language Models・Semantic Text Processing
- deberta-v2-base-japaneseku-nlp24kJapanese DeBERTa V2 base model pretrained on 171 GB of Japanese Wikipedia, CC‑100 and OSCAR data using Juman++ segmentation and SentencePiece tokenization, trained for three weeks on eight NVIDIA A100 GPUs and ready for fine‑tuning.HF modelShared: Language Models・Semantic Text Processing
- modernbert-ja-130msbintuitions12kA 132‑million‑parameter Japanese ModernBERT model that blends local‑global and RoPE attention, trained on 4.39 T tokens (Japanese/English) with a 102‑k‑size vocab, 8,192‑token max length, and optimized for Flash Attention 2.HF modelShared: Language Models・Semantic Text Processing
GPTSANtanreinama119General-purpose Swich transformer based Japanese language modeShared: Language Models・Semantic Text Processing