Llama-3-Base-8B-SFT-RRHF: benchmark results
Princeton NLP's Llama-3-Base-8B-SFT-RRHF, a Llama 3 8B base model with SFT and RRHF alignment, released alongside Princeton's SimPO preference-optimization paper. Provider: Princeton NLP. Released 2024-07-06. Access: Open.
Aggregate Rating 1380 ± 21, rank #1282 of 1654 rated models, from 20 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM - Ko-GPQA-Diamond | 27.27 | Normalized accuracy (%) | 92.4 |
| Open Korean LLM - KorNAT-Harmless | 65.44 | Normalized accuracy (%) | 80.3 |
| Open Korean LLM - Ko-Winogrande | 63.3 | Accuracy (%) | 73.4 |
| Open Korean LLM - Ko-EQ-Bench | 20.37 | EQ-Bench score | 66.3 |
| Open Portuguese LLM - ENEM | 66.13 | Accuracy (%) | 63.6 |
| Open Korean LLM Leaderboard | 37.57 | Average Score (%) | 62.6 |
| Open Portuguese LLM - OAB Exams | 45.97 | Accuracy (%) | 62.6 |
| Open LLM Leaderboard - GPQA | 30.54 | Score | 62.3 |
| Open Korean LLM - KorNAT-Helpful | 47.8 | Normalized accuracy (%) | 60.8 |
| Open Portuguese LLM - BLUEX | 53.82 | Accuracy (%) | 57.4 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 14.25 | Exact match, flexible extract (%) | 42.6 |
| Open Korean LLM - Ko-IFEval | 25.93 | Strict accuracy, prompt/instruction mean (%) | 37.3 |
Interactive version: theaggregate.ai/model?slug=llama-3-base-8b-sft-rrhf · How It Works · Data refreshed daily, snapshot 2026-10-24.