Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Visual Question Answering on VQAv2 3K curated MS COCO (test)
Loading...
7.4
Relative Performance Drop (%)
DLA
6.236
14.093
21.95
29.807
Apr 20, 2026
Relative Performance Drop (%)
Updated 1mo ago
Evaluation Results
Method
Method
Links
Relative Performance Drop (%)
DLA
Base Model=LLaVA-1.5-7...
2026.04
7.4
DLA
Base Model=Qwen2.5-VL-...
2026.04
9.8
LLM-Knowledge
Base Model=LLaVA-1.5-7...
2026.04
16.5
Group Patching
Base Model=LLaVA-1.5-7...
2026.04
18
LLM-Knowledge
Base Model=Qwen2.5-VL-...
2026.04
19.4
Group Patching
Base Model=Qwen2.5-VL-...
2026.04
20.05
QRNCA
Base Model=LLaVA-1.5-7...
2026.04
20.8
QRNCA
Base Model=Qwen2.5-VL-...
2026.04
24.5
HONES
Base Model=LLaVA-1.5-7...
2026.04
27.3
HONES
Base Model=Qwen2.5-VL-...
2026.04
36.5
Feedback
Search any
task
Search any
task