llama2-ko-13B-instruct-v1.2: benchmark results
Provider: Meta. Access: Open.
Unified ELO 1303 ± 18, rank #2449 of 2656 rated models, from 10 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM - Ko-GSM8K (flexible extract) | 18.88 | Exact match, flexible extract (%) | 50.1 |
| Open Korean LLM - KorNAT-CKA | 21.3 | Normalized accuracy (%) | 34 |
| Open Korean LLM - Ko-Winogrande | 57.54 | Accuracy (%) | 32.2 |
| Open Korean LLM - KorNAT-Harmless | 61.47 | Normalized accuracy (%) | 29.2 |
| Open Korean LLM - Ko-EQ-Bench | 0 | EQ-Bench score | 24.5 |
| Open Korean LLM - Ko-GPQA-Diamond | 19.19 | Normalized accuracy (%) | 6.4 |
| Open Korean LLM - KorNAT-Helpful | 39.95 | Normalized accuracy (%) | 6.2 |
| Open Ko-LLM Leaderboard | 28.14 | Average (%) | 5.1 |
| Open Korean LLM Leaderboard | 28.14 | Average Score (%) | 5.1 |
| Open Korean LLM - Ko-IFEval | 2.9 | Strict accuracy, prompt/instruction mean (%) | 4.8 |
Interactive version: theaggregate.ai/model?slug=llama2-ko-13b-instruct-v1-2 · How It Works · Data refreshed daily, snapshot 2026-09-19.