llama-2-koen-13B-sft-v0.3: benchmark results
Provider: Meta. Access: Open.
Unified ELO 1351 ± 18, rank #2230 of 2656 rated models, from 10 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM - Ko-Winogrande | 66.61 | Accuracy (%) | 88.8 |
| Open Korean LLM - KorNAT-Harmless | 66.43 | Normalized accuracy (%) | 86 |
| Open Korean LLM - KorNAT-Helpful | 50.25 | Normalized accuracy (%) | 84.8 |
| Open Korean LLM - KorNAT-CKA | 20.94 | Normalized accuracy (%) | 32 |
| Open Korean LLM - Ko-EQ-Bench | 0 | EQ-Bench score | 24.5 |
| Open Korean LLM Leaderboard | 30.58 | Average Score (%) | 20.2 |
| Open Ko-LLM Leaderboard | 30.58 | Average (%) | 20 |
| Open Korean LLM - Ko-GPQA-Diamond | 20.2 | Normalized accuracy (%) | 13.9 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 0.53 | Exact match, flexible extract (%) | 9 |
| Open Korean LLM - Ko-IFEval | 0.18 | Strict accuracy, prompt/instruction mean (%) | 1.5 |
Interactive version: theaggregate.ai/model?slug=llama-2-koen-13b-sft-v0-3 · How It Works · Data refreshed daily, snapshot 2026-09-19.