Llama-3-Instruct-8B-CPO: benchmark results
Princeton NLP's Llama-3-Instruct-8B-CPO, a Llama 3 8B Instruct model aligned via CPO, released alongside Princeton's SimPO preference-optimization paper. Provider: Princeton NLP. Released 2024-07-06. Access: Open.
Aggregate Rating 1443 ± 19, rank #1038 of 1661 rated models, from 21 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM - Ko-EQ-Bench | 41.07 | EQ-Bench score | 91.9 |
| Open Korean LLM - Ko-IFEval | 45.4 | Strict accuracy, prompt/instruction mean (%) | 89.5 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 53.9 | Exact match, flexible extract (%) | 87.7 |
| Open LLM Leaderboard - IFEval | 72.93 | Score | 87 |
| Open Portuguese LLM - FaQuAD NLI | 77.26 | Macro F1 (%) | 79 |
| Open Korean LLM Leaderboard | 40.66 | Average Score (%) | 78.3 |
| Open Ko-LLM Leaderboard | 40.66 | Average (%) | 78.2 |
| Open Korean LLM - KorNAT-Helpful | 49 | Normalized accuracy (%) | 75.6 |
| Open Portuguese LLM - OAB Exams | 49.7 | Accuracy (%) | 74.2 |
| Open Portuguese LLM - ENEM | 69.84 | Accuracy (%) | 74.1 |
| Open Portuguese LLM - BLUEX | 57.72 | Accuracy (%) | 70.7 |
| Open Korean LLM - Ko-GPQA-Diamond | 24.75 | Normalized accuracy (%) | 69.5 |
Interactive version: theaggregate.ai/model?slug=llama-3-instruct-8b-cpo · How It Works · Data refreshed daily, snapshot 2026-10-23.