A Japanese-optimized large language model developed by NII with 13 billion parameters. It was trained on 2.1 trillion tokens from the llm-jp-corpus v3 dataset.
Parameters
Undisclosed
Context Window
License
Proprietary
Release Date
2024-04-23
API Pricing
API pricing for this model is not yet available
Strengths
- •Trained on a massive, high-quality Japanese corpus.
- •13B parameter size offers a strong balance of performance and efficiency.
- •Solid foundation for Japanese language tasks.
Weaknesses
- •Not instruction-tuned by default; requires fine-tuning for chat.
- •Lacks the mixture-of-experts architecture found in newer models.
- •Performance may be surpassed by the largest llm-jp-3 models.
Use Cases
- •Base model for further pre-training on domain-specific Japanese text.
- •A starting point for creating instruction-tuned Japanese models.
- •Academic research on Japanese language model training.