llm-jp-3-3.7B-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 1440 ± 52, rank #4207 of 10916 rated models, from 62 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Japanese LLM - Xlsum JA Rouge1 | 35.87 | Score (%) | 95.4 |
| Open Japanese LLM - SUM | 12.83 | Score (%) | 94.7 |
| Open Japanese LLM - Xlsum JA Rouge2 | 12.81 | Score (%) | 94.7 |
| Open Japanese LLM - Xlsum JA RougeLsum | 29.37 | Score (%) | 94.7 |
| Open Japanese LLM - Xlsum JA Bert Score JA F1 | 71.19 | Score (%) | 93 |
| Open Japanese LLM - AIO Char F1 | 71.41 | Score (%) | 83.9 |
| Open Japanese LLM - Xlsum JA Bleu JA | 3 | Score (%) | 80.8 |
| pfgen-bench - Completion Mode - Fluency | 0.74 | Fluency Score | 79.6 |
| pfgen-bench - Completion Mode - Score | 0.63 | pfgen Score (mean of three) | 78.6 |
| pfgen-bench - Completion Mode - Truthfulness | 0.86 | Truthfulness Score | 78.4 |
| pfgen-bench - Completion Mode - Helpfulness | 0.29 | Helpfulness Score | 77.5 |
| Open Japanese LLM - QA | 57.09 | Score (%) | 76.7 |
Interactive version: theaggregate.ai/model?slug=llm-jp-3-3-7b-instruct · How It Works · Data refreshed daily, snapshot 2026-11-08.