Llama-3-Base-8B-SFT-IPO: benchmark results
Princeton NLP's Llama-3-Base-8B-SFT-IPO, a Llama 3 8B base model with SFT and IPO alignment, released alongside Princeton's SimPO preference-optimization paper. Provider: Princeton NLP. Released 2024-05-17. Access: Open.
Aggregate Rating 1412 ± 21, rank #1176 of 1654 rated models, from 20 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM - KorNAT-Helpful | 54.78 | Normalized accuracy (%) | 95.5 |
| Open Korean LLM - Ko-EQ-Bench | 30 | EQ-Bench score | 78.5 |
| Open Korean LLM - KorNAT-Harmless | 64.29 | Normalized accuracy (%) | 72.3 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 37.3 | Exact match, flexible extract (%) | 71.2 |
| Open Korean LLM Leaderboard | 39.05 | Average Score (%) | 69.4 |
| Open Portuguese LLM - ASSIN2 RTE | 91.62 | Macro F1 (%) | 68.8 |
| Open Portuguese LLM - OAB Exams | 46.38 | Accuracy (%) | 63.7 |
| Open Portuguese LLM - ENEM | 65.15 | Accuracy (%) | 60.6 |
| Open Korean LLM - Ko-GPQA-Diamond | 23.74 | Normalized accuracy (%) | 55.6 |
| Open Korean LLM - Ko-Winogrande | 60.46 | Accuracy (%) | 54.5 |
| Open Portuguese LLM - BLUEX | 52.71 | Accuracy (%) | 53.8 |
| Open LLM Leaderboard - GPQA | 29.78 | Score | 53.6 |
Interactive version: theaggregate.ai/model?slug=llama-3-base-8b-sft-ipo · How It Works · Data refreshed daily, snapshot 2026-10-24.