Factual Knowledge Evaluation on WikiBench
28.75WikiBench ScoreGenius
Evaluation Results
| Method | Links | |
|---|---|---|
| GeniusTraining Data Source=Magpie, Training Sample Size=25K2025.04 | 28.75 | |
| LLaMA3.1-8B-Instruct2025.04 | 27.65 | |
| GeniusTraining Data Source=OpenHermes, Training Sample Size=32K2025.04 | 27 |