← all datasets

OmniAlign-V

Emerging
1papers using it
2025first seen

OmniAlign-V is a dataset containing 200K high-quality training samples of diverse images, complex questions, and varied response formats, used to improve the alignment of multi-modal large language models (MLLMs) with human preferences.

🔎 Find this dataset

Papers using OmniAlign-V (1)

OmniAlign-V dataset — papers, benchmarks & downloads · Large Language Models