Small Model Plus Big Model Team Up to Catch AI Image Mistakes

Original: Two-Token Features and Small-Large Ensembles for VLM Hallucination Detection

Source: arxiv.org ↗