Llama-3-Base-8B-SFT-RDPO: benchmark results
Princeton NLP's R-DPO baseline trained on their Llama 3 8B SFT base model, from the SimPO paper's base-model comparison setup. Provider: Princeton NLP. Released 2024-05-17. Access: Open.
Aggregate Rating 1454 ± 22, rank #975 of 1654 rated models, from 20 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM - KorNAT-Helpful | 56.23 | Normalized accuracy (%) | 96.2 |
| Open Korean LLM - Ko-EQ-Bench | 44.31 | EQ-Bench score | 93.4 |
| Open Korean LLM - KorNAT-Harmless | 67.37 | Normalized accuracy (%) | 89.2 |
| Open Korean LLM Leaderboard | 43.44 | Average Score (%) | 88 |
| Open Korean LLM - Ko-GPQA-Diamond | 25.25 | Normalized accuracy (%) | 75.9 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 38.59 | Exact match, flexible extract (%) | 73 |
| Open Portuguese LLM - ASSIN2 RTE | 91.58 | Macro F1 (%) | 68.4 |
| Open Portuguese LLM - OAB Exams | 46.51 | Accuracy (%) | 64 |
| Open LLM Leaderboard - GPQA | 30.62 | Score | 63.2 |
| Open Portuguese LLM - ENEM | 65.71 | Accuracy (%) | 62.5 |
| Open Korean LLM - Ko-IFEval | 31.12 | Strict accuracy, prompt/instruction mean (%) | 59 |
| Open Portuguese LLM - BLUEX | 53.96 | Accuracy (%) | 58 |
Interactive version: theaggregate.ai/model?slug=llama-3-base-8b-sft-rdpo · How It Works · Data refreshed daily, snapshot 2026-10-24.