A Japanese-optimized large language model developed by NII. It is one of eight model sizes ranging from 150M to 172B parameters. Trained on the llm-jp-corpus v3 dataset, with SFT and DPO applied via Instruct2/Instruct3. Models under 13B are released under Apache 2.0 license.
Parameters
Undisclosed
Context Window
License
Proprietary
Release Date
2024-09-29
API Pricing
API pricing for this model is not yet available
Strengths
- •Largest available size (172B) offers peak performance potential.
- •Excellently optimized for Japanese language nuances.
- •Underwent advanced alignment processes (SFT/DPO).
Weaknesses
- •Requires immense computational power for training and inference.
- •Extremely large file size makes deployment challenging.
- •May produce overly verbose or redundant outputs without careful prompting.
Use Cases
- •Research benchmarking for state-of-the-art Japanese LLMs.
- •Tasks demanding the highest level of language comprehension.
- •Foundation for training even more specialized or larger models.