Llama-3-Base-8B-SFT-SimPO: benchmark results
Provider: Princeton NLP. Released 2024-05-24. Access: Open.
Aggregate Rating 1418 ± 20, rank #1160 of 1661 rated models, from 21 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM - KorNAT-Helpful | 53.5 | Normalized accuracy (%) | 94.4 |
| Open Korean LLM - KorNAT-Harmless | 66.09 | Normalized accuracy (%) | 84.2 |
| Open Korean LLM - Ko-GPQA-Diamond | 25.76 | Normalized accuracy (%) | 81.3 |
| Open Korean LLM - Ko-IFEval | 35.87 | Strict accuracy, prompt/instruction mean (%) | 74.2 |
| Open Korean LLM Leaderboard | 39.75 | Average Score (%) | 73.9 |
| Open Ko-LLM Leaderboard | 39.75 | Average (%) | 73.7 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 32.83 | Exact match, flexible extract (%) | 66.6 |
| Open Portuguese LLM - ASSIN2 RTE | 91.25 | Macro F1 (%) | 65.6 |
| Open Korean LLM - Ko-Winogrande | 61.96 | Accuracy (%) | 65.3 |
| Open Portuguese LLM - OAB Exams | 46.88 | Accuracy (%) | 64.7 |
| Open Portuguese LLM - ENEM | 65.36 | Accuracy (%) | 61.4 |
| Open Korean LLM - KorNAT-CKA | 25.5 | Normalized accuracy (%) | 58.6 |
Interactive version: theaggregate.ai/model?slug=llama-3-base-8b-sft-simpo · How It Works · Data refreshed daily, snapshot 2026-10-23.