Mistral-7B-Instruct-RRHF: benchmark results
Princeton NLP's Mistral-7B-Instruct-RRHF, a Mistral 7B Instruct model aligned via RRHF, released alongside Princeton's SimPO preference-optimization paper. Provider: Princeton NLP. Released 2024-07-06. Access: Open.
Aggregate Rating 1387 ± 19, rank #1270 of 1661 rated models, from 21 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM - KorNAT-Helpful | 51.53 | Normalized accuracy (%) | 91 |
| Open Korean LLM - KorNAT-Harmless | 67.3 | Normalized accuracy (%) | 88.8 |
| Open Korean LLM - Ko-IFEval | 35.93 | Strict accuracy, prompt/instruction mean (%) | 74.4 |
| Open Korean LLM - Ko-EQ-Bench | 23.67 | EQ-Bench score | 69.1 |
| Open Portuguese LLM - FaQuAD NLI | 69.41 | Macro F1 (%) | 58.9 |
| Open LLM Leaderboard - IFEval | 49.6 | Score | 58.2 |
| Open Ko-LLM Leaderboard | 36.6 | Average (%) | 58.1 |
| Open Korean LLM Leaderboard | 36.6 | Average Score (%) | 58.1 |
| Open Portuguese LLM - ASSIN2 RTE | 90.28 | Macro F1 (%) | 55.8 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 22.06 | Exact match, flexible extract (%) | 54.3 |
| Open Portuguese LLM - BLUEX | 51.88 | Accuracy (%) | 50.8 |
| Open Portuguese LLM - ENEM | 58.15 | Accuracy (%) | 43.3 |
Interactive version: theaggregate.ai/model?slug=mistral-7b-instruct-rrhf · How It Works · Data refreshed daily, snapshot 2026-10-23.