A new tool to evaluate biases in large vision-language models.
― 6 min read
Shiguang Shan
Dysca introduces a new way to assess LVLM performance using synthetic data.
― 6 min read
A new defense method to enhance safety in text-to-image diffusion models.
― 5 min read
UniPose integrates multiple methods for better human pose understanding and generation.
― 7 min read