llama2-ko-chang-instruct-chat: benchmark results
Provider: Meta. Access: Open.
Unified ELO 1309 ± 17, rank #2427 of 2656 rated models, from 10 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM - KorNAT-Helpful | 46.8 | Normalized accuracy (%) | 49.4 |
| Open Korean LLM - KorNAT-CKA | 23.1 | Normalized accuracy (%) | 42.6 |
| Open Korean LLM - Ko-Winogrande | 57.62 | Accuracy (%) | 32.9 |
| Open Korean LLM - Ko-EQ-Bench | 0 | EQ-Bench score | 24.5 |
| Open Korean LLM - KorNAT-Harmless | 60.82 | Normalized accuracy (%) | 19.5 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 2.35 | Exact match, flexible extract (%) | 18.9 |
| Open Korean LLM - Ko-GPQA-Diamond | 20.71 | Normalized accuracy (%) | 18.8 |
| Open Korean LLM - Ko-IFEval | 8.45 | Strict accuracy, prompt/instruction mean (%) | 7.3 |
| Open Ko-LLM Leaderboard | 28.43 | Average (%) | 6.3 |
| Open Korean LLM Leaderboard | 28.43 | Average Score (%) | 6.3 |
Interactive version: theaggregate.ai/model?slug=llama2-ko-chang-instruct-chat · How It Works · Data refreshed daily, snapshot 2026-09-19.