Mitigation Methods - Uncovers object hallucinations via logical closed loops.
[ACL 2024] Logical Closed Loop: Uncovering Object Hallucinations in Large Vision-Language Models. Detect and mitigate object hallucinations in LVLMs by itself through logical closed loops.
