sarashina2-7B: benchmark results
Provider: Other. Access: Open.
Unified ELO 1439 ± 45, rank #1595 of 2656 rated models, from 26 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| pfgen-bench - Completion Mode - Score | 0.65 | pfgen Score (mean of three) | 82.1 |
| pfgen-bench - Completion Mode - Truthfulness | 0.87 | Truthfulness Score | 82.1 |
| pfgen-bench - Completion Mode - Helpfulness | 0.32 | Helpfulness Score | 81.5 |
| pfgen-bench - Completion Mode - Fluency | 0.75 | Fluency Score | 80.4 |
| Swallow - Pre-trained Japanese - NIILC | 63.4 | Character F1 (%) | 77.6 |
| Swallow - Pre-trained Japanese - WMT20 En-Ja | 27.3 | BLEU | 67.3 |
| Swallow - Pre-trained English - XWINO | 89.2 | Accuracy (%) | 57.1 |
| Swallow - Pre-trained Japanese - JEMHopQA | 50.9 | Character F1 (%) | 54.1 |
| Swallow - Pre-trained Japanese - WMT20 Ja-En | 20.1 | BLEU | 49 |
| Swallow - Pre-trained English - OpenBookQA | 34.6 | Accuracy (%) | 43.9 |
| Swallow - Pre-trained Japanese - JSQuAD | 86.8 | Character F1 (%) | 40.8 |
| Swallow - Pre-trained Japanese - Average | 39.5 | Average Score (%) | 38.8 |
Interactive version: theaggregate.ai/model?slug=sarashina2-7b · How It Works · Data refreshed daily, snapshot 2026-09-19.