
Link to the Low-Rank Adapter created by fine-tuning LLaMA using the Stanford Alpaca dataset translated into Japanese, and sample code for generating it.
Description in other languages
Statistics
Statistics as of Nov 17, 2026
GitHub stars
141
±0 (Oct 13〜)
Last commit
Apr 2023
Created on GitHub
Mar 22, 2023
Listed
Mar 27, 2023
Stars over time
Show as a table
| Month | Stars |
|---|---|
| Dec 2025 | 141 |
| Jan 2026 | 141 |
| Feb 2026 | 141 |
| Mar 2026 | 141 |
| Apr 2026 | 141 |
| May 2026 | 141 |
| Jun 2026 | 141 |
| Jul 2026 | 141 |
| Aug 2026 | 141 |
| Sep 2026 | 141 |
| Oct 2026 | 141 |
| Nov 2026 | 141 |
Last snapshot of each month. Star counts of 1,000 or more are rounded to the nearest 100.
Popular in the same category
lukestudio-ousia727LUKE -- Language Understanding with Knowledge-based Embeddings
bert-japanesecl-tohoku551BERT models for Japanese language text.
bert-japaneseyoheikikuta497BERT with SentencePiece for Japanese text.
gpt2-japanesetanreinama322Japanese GPT2 Generation Model
japanese-dialog-transformersnttcslab247Code for evaluating Japanese pretrained models provided by NTT Ltd.
DistilBERT-base-jpBandaiNamcoResearchInc161A Japanese DistilBERT pretrained model, which was trained on Wikipedia.
Resources in similar research fields
- deberta-v2-base-japanese-char-wwmku-nlp24kA Japanese DeBERTa‑V2 base model pre‑trained on 171 GB of Japanese Wikipedia, CC‑100, and OSCAR text with character‑level tokenization, whole‑word masking, trained for 20 days on 8 A100 GPUs, and ready for downstream fine‑tuning.HF modelShared: Language Models・Representation Learning
- modernbert-ja-30msbintuitions3kModernBERT‑Ja‑30M is a Japanese‑language BERT variant that blends local and global attention with RoPE, trained on 4.39 TB of Japanese/English text, supports 8,192‑token sequences, comes in sizes from 30 M to 130 M parameters, and works best with Flash Attention 2.HF modelShared: Language Models・Representation Learning
- modernbert-ja-70msbintuitions2kModernBERT‑Ja‑70M is a lightweight Japanese BERT variant that blends local and global attention with RoPE, trained on 4.39 T mixed‑language tokens (vocab 102 400, max 8 192 tokens), supports Flash Attention 2, and comes in several sizes from 30 M to 310 M parameters.HF modelShared: Language Models・Representation Learning
- bert-large-japanese-v2tohoku-nlp66kJapanese‑BERT‑Large trained on CC‑100 and Wikipedia, using Unidic‑lite word‑level tokenization with WordPiece subwords and whole‑word masking (24 layers, 1024‑dim hidden, 16 heads, 32k vocab), with pretraining code on cl‑tohoku/bert‑japanese.HF modelShared: Representation Learning・Language Models
- bert-base-japanese-v3tohoku-nlp66kJapanese BERT‑base (12 layers, 768‑dim hidden, 12 heads, 32 k vocab) pretrained with whole‑word masking on CC‑100 and 2023‑Jan Wikipedia, using Unidic 2.1.2 word‑level tokenization plus WordPiece, in 2 M training steps.HF modelShared: Representation Learning・Language Models
Laboro-BERT-Japaneselaboroai73Laboro BERT Japanese: Japanese BERT Pre-Trained With Web-CorpusShared: Representation Learning・Language Models