llm-jp-3-13B: benchmark results
National Institute of Informatics' llm-jp-3-13b, a base transformer language model pretrained on 2.1T tokens, part of the LLM-jp-3 series. Provider: National Institute of Informatics. Released 2024-09-23. Access: Open.
Aggregate Rating 1403 ± 31, rank #1203 of 1635 rated models, from 76 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Japanese LLM - AIO Char F1 | 85.77 | Score (%) | 94.7 |
| Open Japanese LLM - Niilc Char F1 | 64.24 | Score (%) | 91.8 |
| Open Japanese LLM - QA | 67.21 | Score (%) | 89.9 |
| Open Japanese LLM - Wiki Reading Char F1 | 86.93 | Score (%) | 89.4 |
| Open Japanese LLM - Mbpp Pylint Check | 95.78 | Score (%) | 87 |
| Open Japanese LLM - ALT E TO J Bert Score JA F1 | 86.82 | Score (%) | 86.3 |
| Open Japanese LLM - ALT E TO J Bleu JA | 13.53 | Score (%) | 84.5 |
| pfgen-bench - Completion Mode - Truthfulness | 0.88 | Truthfulness Score | 84.2 |
| pfgen-bench - Completion Mode - Fluency | 0.77 | Fluency Score | 84 |
| pfgen-bench - Completion Mode - Helpfulness | 0.35 | Helpfulness Score | 84 |
| Swallow - Pre-trained Japanese - NIILC | 64.9 | Character F1 (%) | 83.7 |
| pfgen-bench - Completion Mode - Score | 0.67 | pfgen Score (mean of three) | 83.7 |
Interactive version: theaggregate.ai/model?slug=llm-jp-3-13b · How It Works · Data refreshed daily, snapshot 2026-10-15.