EEVE-Korean-10.8B-instruct-v4-sft: benchmark results
Provider: Yanolja. Released 2024-07-02. Access: Open.
Aggregate Rating 1438 ± 19, rank #1059 of 1656 rated models, from 19 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM - KorNAT-CKA | 35.62 | Normalized accuracy (%) | 95.6 |
| Open Korean LLM - Ko-Winogrande | 66.69 | Accuracy (%) | 89.2 |
| Open Korean LLM Leaderboard | 43.61 | Average Score (%) | 88.3 |
| Open Ko-LLM Leaderboard | 43.61 | Average (%) | 88.2 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 49.05 | Exact match, flexible extract (%) | 82.5 |
| Open Korean LLM - Ko-EQ-Bench | 27.99 | EQ-Bench score | 75.3 |
| Open Korean LLM - Ko-IFEval | 34.79 | Strict accuracy, prompt/instruction mean (%) | 71.3 |
| Open Korean LLM - Ko-GPQA-Diamond | 23.74 | Normalized accuracy (%) | 55.6 |
| LogicKor - Grammar | 4 | Score (0-10) | 31.5 |
| LogicKor - Coding | 5.85 | Score (0-10) | 26.5 |
| Open Korean LLM - KorNAT-Harmless | 61.16 | Normalized accuracy (%) | 24.2 |
| Open Korean LLM - KorNAT-Helpful | 43.03 | Normalized accuracy (%) | 21.1 |
Interactive version: theaggregate.ai/model?slug=eeve-korean-10-8b-instruct-v4-sft · How It Works · Data refreshed daily, snapshot 2026-10-20.