zephyr-7B-sft-full-orpo: benchmark results
Provider: HuggingFace. Access: Open.
Aggregate Rating 1358 ± 27, rank #7498 of 10916 rated models, from 15 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - WinoGrande | 78.06 | Accuracy (%) (5-shot) | 60.4 |
| Open Korean LLM - KorNAT-Harmless | 63.21 | Normalized accuracy (%) | 58 |
| Open Korean LLM - Ko-IFEval | 30.58 | Strict accuracy, prompt/instruction mean (%) | 57 |
| Open Korean LLM - KorNAT-Helpful | 47.23 | Normalized accuracy (%) | 54.2 |
| Open Korean LLM - Ko-EQ-Bench | 5.15 | EQ-Bench score | 50.8 |
| Open Korean LLM - Ko-Winogrande | 59.98 | Accuracy (%) | 50.5 |
| Open LLM Leaderboard v1 - HellaSwag | 81.89 | Normalized accuracy (%) (10-shot) | 47.5 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 16.45 | Exact match, flexible extract (%) | 46.9 |
| Open LLM Leaderboard v1 - MMLU | 58.13 | Accuracy (%) (5-shot) | 42.6 |
| Open Korean LLM Leaderboard | 32.94 | Average Score (%) | 36.6 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 46.37 | MC2 (%) (0-shot) | 34.2 |
| Open LLM Leaderboard v1 - ARC Challenge | 54.52 | Normalized accuracy (%) (25-shot) | 31.6 |
Interactive version: theaggregate.ai/model?slug=zephyr-7b-sft-full-orpo · How It Works · Data refreshed daily, snapshot 2026-11-08.