llm-jp-3-13B-instruct: benchmark results
LLM-jp 3 instruction-tuned release developed by NII's LLM-jp project, using Llama architecture and a 4,096-token context window. The named instruction-tuning generation remains distinct from the pretrained checkpoint and other instruct generations. Provider: National Institute of Informatics. Access: Open.
Aggregate Rating 1531 ± 33, rank #1940 of 10916 rated models, from 62 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| pfgen-bench - Completion Mode - Helpfulness | 0.44 | Helpfulness Score | 93 |
| Open Japanese LLM - AIO Char F1 | 84.15 | Score (%) | 92.5 |
| pfgen-bench - Completion Mode - Score | 0.72 | pfgen Score (mean of three) | 91.7 |
| Open Japanese LLM - Niilc Char F1 | 64.03 | Score (%) | 91.6 |
| pfgen-bench - Completion Mode - Truthfulness | 0.91 | Truthfulness Score | 90.4 |
| Open Japanese LLM - Wiki Reading Char F1 | 86.65 | Score (%) | 88.7 |
| Open Japanese LLM - Xlsum JA Rouge2 | 11.45 | Score (%) | 88.3 |
| Open Japanese LLM - SUM | 11.43 | Score (%) | 88.2 |
| pfgen-bench - Completion Mode - Fluency | 0.81 | Fluency Score | 87.9 |
| Open Japanese LLM - Wiki Dependency SET F1 | 40.27 | Score (%) | 85.1 |
| pfgen-bench - QA Mode - Fluency | 0.78 | Fluency Score | 84.5 |
| pfgen-bench - QA Mode - Helpfulness | 0.43 | Helpfulness Score | 84.5 |
Interactive version: theaggregate.ai/model?slug=llm-jp-3-13b-instruct · How It Works · Data refreshed daily, snapshot 2026-11-08.