llm-jp-3-980m: 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 1247 ± 44, rank #9277 of 10916 rated models, from 54 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Japanese LLM - AIO Char F1 | 48.8 | Score (%) | 38.9 |
| Open Japanese LLM - Mbpp Pylint Check | 9.04 | Score (%) | 38.7 |
| pfgen-bench - Completion Mode - Truthfulness | 0.63 | Truthfulness Score | 32.3 |
| pfgen-bench - Completion Mode - Score | 0.38 | pfgen Score (mean of three) | 28.4 |
| pfgen-bench - Completion Mode - Fluency | 0.46 | Fluency Score | 26.5 |
| Open Japanese LLM - QA | 39.09 | Score (%) | 25.5 |
| pfgen-bench - Completion Mode - Helpfulness | 0.04 | Helpfulness Score | 23.8 |
| Open Japanese LLM - Niilc Char F1 | 34.35 | Score (%) | 23.1 |
| Open Japanese LLM - Wiki Reading Char F1 | 49.45 | Score (%) | 22.8 |
| Open Japanese LLM - Jsick Exact Match | 61.74 | Score (%) | 22.6 |
| Open Japanese LLM - Wiki Dependency SET F1 | 20.04 | Score (%) | 22 |
| Open Japanese LLM - FA | 14.06 | Score (%) | 19.5 |
Interactive version: theaggregate.ai/model?slug=llm-jp-3-980m · How It Works · Data refreshed daily, snapshot 2026-11-08.