ELYZA-japanese-Llama-2-7B-fast-instruct: benchmark results
Provider: Other. Access: Open.
Unified ELO 1360 ± 20, rank #2414 of 2928 rated models, from 14 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| pfgen-bench - Completion Mode - Helpfulness | 0.08 | Helpfulness Score | 38.2 |
| pfgen-bench - Completion Mode - Truthfulness | 0.64 | Truthfulness Score | 36.9 |
| pfgen-bench - Completion Mode - Score | 0.41 | pfgen Score (mean of three) | 36.4 |
| pfgen-bench - Completion Mode - Fluency | 0.5 | Fluency Score | 35 |
| Open LLM Leaderboard v1 - HellaSwag | 77.55 | Normalized accuracy (%) (10-shot) | 31.1 |
| Open LLM Leaderboard v1 - ARC Challenge | 53.75 | Normalized accuracy (%) (25-shot) | 30.2 |
| Open LLM Leaderboard v1 - WinoGrande | 71.59 | Accuracy (%) (5-shot) | 26.9 |
| Open LLM Leaderboard v1 - MMLU | 46.85 | Accuracy (%) (5-shot) | 25.7 |
| Open LLM Leaderboard v1 - GSM8K | 6.29 | Accuracy (%) (5-shot) | 25.4 |
| pfgen-bench - QA Mode - Helpfulness | 0.06 | Helpfulness Score | 23 |
| pfgen-bench - QA Mode - Score | 0.36 | pfgen Score (mean of three) | 21.2 |
| pfgen-bench - QA Mode - Fluency | 0.44 | Fluency Score | 20 |
Interactive version: theaggregate.ai/model?slug=elyza-japanese-llama-2-7b-fast-instruct · How It Works · Data refreshed daily, snapshot 2026-09-23.