EEVE-Korean-Instruct-10.8B-v1.0: benchmark results
Yanolja's 10.8B Korean vocabulary-extended SOLAR-10.7B (February 2024), DPO instruction-tuned for Korean and English. Provider: Yanolja. Released 2024-02-22. Access: Open.
Aggregate Rating 1477 ± 17, rank #859 of 1656 rated models, from 25 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM - Ko-EQ-Bench | 51.67 | EQ-Bench score | 95.6 |
| Open Korean LLM - Ko-Winogrande | 69.77 | Accuracy (%) | 95.6 |
| Open Korean LLM Leaderboard | 47.08 | Average Score (%) | 94.4 |
| Open Ko-LLM Leaderboard | 47.08 | Average (%) | 94.3 |
| Open Korean LLM - KorNAT-CKA | 33.56 | Normalized accuracy (%) | 92.3 |
| Open Korean LLM - KorNAT-Harmless | 65.72 | Normalized accuracy (%) | 81.9 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 47.61 | Exact match, flexible extract (%) | 81.5 |
| Open LLM Leaderboard v1 - WinoGrande | 81.93 | Accuracy (%) (5-shot) | 81.3 |
| Open Korean LLM - Ko-IFEval | 35.81 | Strict accuracy, prompt/instruction mean (%) | 74.1 |
| Open Korean LLM - KorNAT-Helpful | 48.48 | Normalized accuracy (%) | 68.6 |
| Open LLM Leaderboard v1 - ARC Challenge | 64.85 | Normalized accuracy (%) (25-shot) | 68.1 |
| Open LLM Leaderboard v1 - MMLU | 64.23 | Accuracy (%) (5-shot) | 67 |
Interactive version: theaggregate.ai/model?slug=eeve-korean-instruct-10-8b-v1-0 · How It Works · Data refreshed daily, snapshot 2026-10-20.