Our new X account is live! Follow @wizwand_team for updates

Visual Question Answering on InfoSeek

55.93Accuracy

MAD-RAG

Updated 3d ago

Evaluation Results

Method	Links
MAD-RAG 2026.01		55.93
MAD-RAG 2026.01		53.29
Vanilla RAG 2026.01		52.66
Vanilla RAG 2026.01		52.12
GoG-2.5-7B-RL 2026.01		51.05
GoG-3-8B-Think-RL 2026.01		49.05
GoG-2.5-7B-SFT 2026.01		41.9
GoG-3-8B-Think-SFT 2026.01		40.55
Qwen2.5-VL-7B 2026.01		38.15
Qwen3-VL-8B-Thinking 2026.01		32.25
Qwen3-VL-8B-Think 2026.01		32
GPT-4o 2026.01		30.6
MMSearch-R1* 2026.01		24.65
Qwen3-VL-30B-A3B-Thinking 2026.01		22.8
Closed-book 2026.01		18.73
Qwen2.5-VL-7B 2026.01		17.4
Qwen2.5-VL-32B 2026.01		16.95
Qwen3-VL-30B-A3B-Instruct 2026.01		16.7
Qwen3-VL-8B-Thinking 2026.01		16.05
Closed-book 2026.01		15.76
CVLM 2024.02		15.45
Qwen-VL 2024.02		15.21
Qwen2.5-VL-7B 2026.01		13.9
Qwen3-VL-8B-Instruct 2026.01		13.8
CVLM (3M IKPairs, Objects=5) 2024.02		11.27
CVLM (3M IKPairs) w/o FKA 2024.02		11.21
CVLM (3M IKPairs, Objects=8) 2024.02		10.78
CVLM 2024.02		10.72
BLIP2 2024.02		10.67
CVLM (3M IKPairs, Objects=3) 2024.02		10.42
InstructBLIP 2024.02		10.3
InstructBLIP 2024.02		10.05
MiniGPT4 2024.02		10.03
CVLM (3M IKPairs, Objects=1) 2024.02		10.01
CVLM w/o FKA 2024.02		9.33
InstructBLIP 2024.02		8.36
LLaVA-v1.5† 2024.02		8.18
CVLM w/o (FKA & VKA) 2024.02		8.18