Share your thoughts, 1 month free Claude Pro on usSee more

Text-based Visual Question Answering on TextVQA 52

63.8Accuracy

Qwen-VL

Updated 1mo ago

Evaluation Results

Method	Links
Qwen-VL 2024.07		63.8
Qwen-VL-Chat 2024.07		61.5
Bunny-3B 2024.07		60.6
LAF-7B 2024.07		60.1
LLaVA-1.5+ (Ours) 2024.07		59.8
Imp-v1 2024.07		59.4
TinyLLaVA 2024.07		59.1
mPLUG-Owl2 2024.07		58.2
LLaVA-1.5 2024.07		58.2
Mipha-3B+ (Ours) 2024.07		57.8
Mobile VLM-v2-3B 2024.07		57.5
MoE-LLaVA-3.6B 2024.07		57
Mipha-3B 2024.07		56.6
Mobile VLM-v2-1.7B 2024.07		52.1
InstructBLIP 2024.07		50.7
InstructBLIP 2024.07		50.1
LLaVA-Phi 2024.07		48.6
Mobile VLM-3B 2024.07		47.5
BLIP-2 2024.07		42.5
Mobile VLM-1.7B 2024.07		41.5
MC-LLaVA 2024.07		38.6
IDEFICS-80B 2024.07		30.9
IDEFICS-9B 2024.07		25.9