GQA

GQA is a new dataset for real-world visual reasoning and compositional question answering, seeking to address key shortcomings of previous visual question answering (VQA) datasets.

Papers using GQA (4)

K-Search: LLM Kernel Generation via Co-Evolving Intrinsic World Model2026Test-Time Warmup for Multimodal Large Language Models2025ByDeWay: Boost Your multimodal LLM with DEpth prompting in a Training-Free Way2025Enhancing Instruction-Following Capability of Visual-Language Models by Reducing Image Redundancy2024