Llama 3.1 8B Magpie Align V0.1: benchmark results
Provider: Meta. Released 2024-07-24. Access: Open.
Aggregate Rating 1407 ± 45, rank #5731 of 10916 rated models, from 15 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM - KorNAT-Helpful | 55.63 | Normalized accuracy (%) | 96 |
| Open Korean LLM - KorNAT-Harmless | 66.71 | Normalized accuracy (%) | 86.8 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 50.19 | Exact match, flexible extract (%) | 84 |
| Open Korean LLM - Ko-GPQA-Diamond | 25.76 | Normalized accuracy (%) | 81.3 |
| Open Korean LLM Leaderboard | 40.29 | Average Score (%) | 76.2 |
| Open Korean LLM - Ko-Winogrande | 62.59 | Accuracy (%) | 69.2 |
| Open Korean LLM - Ko-IFEval | 33.84 | Strict accuracy, prompt/instruction mean (%) | 69 |
| Open Korean LLM - KorNAT-CKA | 24.27 | Normalized accuracy (%) | 50.7 |
| Open LLM Leaderboard - IFEval | 44.58 | Score | 48.7 |
| Open LLM Leaderboard - MMLU-Pro | 32.62 | Score | 45 |
| Open LLM Leaderboard - BBH | 46.22 | Score | 36.8 |
| Open LLM Leaderboard - MATH Level 5 | 6.65 | Score | 35.9 |
Interactive version: theaggregate.ai/model?slug=llama-3-1-8b-magpie-align-v0-1 · How It Works · Data refreshed daily, snapshot 2026-11-08.