Reference-based captioning on RefCOCOg
65CIDErLLaDA-V + Mask Prior Suppression
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| LLaDA-V + Mask Prior SuppressionConfiguration=With Mask Prior Suppression2026.05 | 65 | 8.4 | 14.6 | 72.8 | |
| LLaDA-VConfiguration=Baseline2026.05 | 64.8 | 8.2 | 14.6 | 72.8 | |
| LaViDa + Mask Prior SuppressionConfiguration=With Mask Prior Suppression2026.05 | 44 | 4.9 | 11.3 | 69.4 | |
| LaViDaConfiguration=Baseline2026.05 | 36.9 | 1.5 | 8.4 | 68.1 |