Be Consistent! Enhancing Robust Visual Reasoning in LVLMs with Consistency Constraints⭐7

New benchmark ConVBench and training method ConVLM enhance logical consistency in vision-language AI, evaluating and improving robust visual reasoning.

Adversarial Style Optimization: Enhancing VLM Jailbreaks by GRPO-Based Stylistic Triggers Optimization⭐7

ASO enhances visual jailbreak attacks on multimodal LLMs by optimizing stylistic triggers via GRPO, achieving higher Attack Success Rates and exposing stylistic...