Hallucination Evaluation on GAVIE
9.33RelevancyCAST
Evaluation results
| Method | Links | ||
|---|---|---|---|
| CASTBackbone=Qwen2.5-VL-7B2026.05 | 9.33 | 8.42 | |
| Qwen2.5-VL-7BBackbone=Qwen2.5-VL-7B2026.05 | 9.2 | 8.09 | |
| CASTBackbone=LLaVA-1.5-7B2026.05 | 8.76 | 6.68 | |
| CASTBackbone=Qwen-VL-Chat2026.05 | 8.63 | 6.6 | |
| Qwen-VL-ChatBackbone=Qwen-VL-Chat2026.05 | 8.26 | 6.39 | |
| LLaVA-1.5-7BBackbone=LLaVA-1.5-7B2026.05 | 8.2 | 6.42 |