Public articles linked to the same research event.
arXiv The work introduces OmniConfess, a training-free method that fixes a candidate response and re-scores it at token resolution under controlled channel-wise evidence interventions, yielding a structured token-by-channel confession that reveals the response's evidential dependence; it uses this confession to preserve grounded content and correct commitments driven by irrelevant or contradictory evidence, and it constructs OmniHalluBench, a 3,540-example benchmark built from six datasets spanning text, image, audio, and video settings and both judgment and free-form generation, with experiments showing hallucination mitigation across heterogeneous modality and task settings.
The work introduces OmniConfess, a training-free method that fixes a candidate response and re-scores it at token resolution under controlled channel-wise evidence interventions, yielding a structured token-by-channel confession that reveals the response's evidential dependence; it uses this confession to preserve grounded content and correct commitments driven by irrelevant or contradictory evidence, and it constructs OmniHalluBench, a 3,540-example benchmark built from six datasets spanning text, image, audio, and video settings and both judgment and free-form generation, with experiments showing hallucination mitigation across heterogeneous modality and task settings.
The work introduces OmniConfess, a training-free method that fixes a candidate response and re-scores it at token resolution under controlled channel-wise evidence interventions, yielding a structured token-by-channel confession that reveals the response's evidential dependence; it uses this confession to preserve grounded content and correct commitments driven by irrelevant or contradictory evidence, and it constructs OmniHalluBench, a 3,540-example benchmark built from six datasets spanning text, image, audio, and video settings and both judgment and free-form generation, with experiments showing hallucination mitigation across heterogeneous modality and task settings.
The work introduces OmniConfess, a training-free method that fixes a candidate response and re-scores it at token resolution under controlled channel-wise evidence interventions, yielding a structured token-by-channel confession that reveals the response's evidential dependence; it uses this confession to preserve grounded content and correct commitments driven by irrelevant or contradictory evidence, and it constructs OmniHalluBench, a 3,540-example benchmark built from six datasets spanning text, image, audio, and video settings and both judgment and free-form generation, with experiments showing hallucination mitigation across heterogeneous modality and task settings.