Jiaxin Ge

PhD student, University of California, Berkeley

1 paper at NeurIPS 2025

Homepage· OpenReview· Semantic Scholar· Google Scholar

Poster Session 4

1 paper

Thursday, December 4, 2025 · 4:30 PM → 7:30 PM

Exhibit Hall C,D,E

Generate, but Verify: Reducing Hallucination in Vision-Language Models with Retrospective Resampling

#4618 · Tsung-Han Wu, Heekyung Lee, Jiaxin Ge, Joseph E. Gonzalez, Trevor Darrell, David M. Chan

We introduce REVERSE, the first framework to integrate generation adjustment with online post-hoc verification within a single VLM architecture. REVERSE detects, backtracks, and corrects hallucinations during the decoding process.