Llama-3-Instruct-8B-CPO-SimPO: benchmark results
Haoran Xu's CPO-SimPO fine-tune of Llama 3 8B Instruct, combining contrastive preference optimization with SimPO's length-normalized objective. Provider: Meta. Released 2024-06-19. Access: Open.
Unified ELO 1496 ± 1, rank #708 of 1392 rated models, from 15 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM Leaderboard | 43.93 | Average Score (%) | 87.7 |
| Open LLM Leaderboard - IFEval | 70.46 | Score | 83.6 |
| Open Chinese LLM - TruthfulQA MC | 55.28 | Accuracy (%) | 62.3 |
| Open Chinese LLM - GSM8K | 47.92 | Accuracy (%) | 60.4 |
| Open LLM Leaderboard - MMLU-Pro | 29.84 | Score | 59.3 |
| Open LLM Leaderboard - BBH | 29.76 | Score | 50.8 |
| Open LLM Leaderboard - MATH Level 5 | 10.27 | Score | 48.3 |
| Open LLM Leaderboard - GPQA | 5.7 | Score | 47.8 |
| Open Chinese LLM - C-Eval Semantic | 68.28 | Accuracy (%) | 45.4 |
| Open Chinese LLM - CMMLU | 52.94 | Accuracy (%) | 45.4 |
| Open Chinese LLM Leaderboard | 55.84 | Average Score (%) | 41.8 |
| Open Chinese LLM - ARC Challenge | 49.66 | Accuracy (%) | 40.2 |
Interactive version: theaggregate.ai/model?slug=llama-3-instruct-8b-cpo-simpo · How It Works · Data refreshed daily, snapshot 2026-09-05.