llm-jp-3-440m: benchmark results
LLM-jp 3 pretrained release developed by NII's LLM-jp project, using Llama architecture and a 4,096-token context window. Provider: National Institute of Informatics. Access: Open.
Aggregate Rating 1232 ± 38, rank #9401 of 10916 rated models, from 54 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Japanese LLM - Mbpp Pylint Check | 33.73 | Score (%) | 51.9 |
| Open Japanese LLM - Wiki NER SET F1 | 1.77 | Score (%) | 27 |
| Open Japanese LLM - AIO Char F1 | 28.93 | Score (%) | 21.2 |
| pfgen-bench - Completion Mode - Truthfulness | 0.54 | Truthfulness Score | 20.8 |
| pfgen-bench - Completion Mode - Helpfulness | 0.03 | Helpfulness Score | 19.5 |
| pfgen-bench - Completion Mode - Score | 0.32 | pfgen Score (mean of three) | 19.2 |
| Open Japanese LLM - Jsem Exact Match | 67.17 | Score (%) | 17.9 |
| pfgen-bench - Completion Mode - Fluency | 0.39 | Fluency Score | 16.3 |
| Open Japanese LLM - QA | 27.26 | Score (%) | 16 |
| Open Japanese LLM - Jsick Exact Match | 55.53 | Score (%) | 15.6 |
| Open Japanese LLM - NLI | 52.96 | Score (%) | 15.4 |
| Open Japanese LLM - Niilc Char F1 | 23.26 | Score (%) | 13.9 |
Interactive version: theaggregate.ai/model?slug=llm-jp-3-440m · How It Works · Data refreshed daily, snapshot 2026-11-08.