Multi-Hop Question Answering on Bamboogle (test val)
45.8Exact Match (EM)DeepControl
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| DeepControlBackbone Model=Qwen2.5-7B2026.02 | 45.8 | 47.9 | |
| DeepControlBackbone Model=Qwen2.5-3B2026.02 | 29.8 | 41.1 |
| Method | Links | ||
|---|---|---|---|
| DeepControlBackbone Model=Qwen2.5-7B2026.02 | 45.8 | 47.9 | |
| DeepControlBackbone Model=Qwen2.5-3B2026.02 | 29.8 | 41.1 |