Visual Question Answering on NewYorker
41.1AccuracyOurs
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| OursBackbone=Qwen2.5-VL-3B2026.01 | 41.1 | 24.1 | 26.6 | |
| GoTVenue=AAAI'24, Backbone=Qwen2.5-VL-3B2026.01 | 40.2 | 23.4 | 28.7 | |
| LADVenue=arXiv'25, Backbone=Qwen2.5-VL-3B2026.01 | 39.4 | 4.4 | 26.8 | |
| CCoTVenue=CVPR'24, Backbone=Qwen2.5-VL-3B2026.01 | 37.9 | 17 | 31.7 | |
| ToTVenue=NIPS'23, Backbone=Qwen2.5-VL-3B2026.01 | 37.1 | 23.4 | 30.6 | |
| CoT-SCVenue=ICLR'23, Backbone=Qwen2.5-VL-3B2026.01 | 35.8 | 20 | 27.8 | |
| DDCoTVenue=NIPS'23, Backbone=Qwen2.5-VL-3B2026.01 | 29.4 | 15.1 | 37 | |
| CoTVenue=NIPS'22, Backbone=Qwen2.5-VL-3B2026.01 | 29.2 | 14.4 | 27.4 |