VDiff-Bench: A Challenging Benchmark for Fine-Grained Image Difference Identification
Published in Arxiv, 2026
We introduce VDiff-Bench, a Benchmark for Fine-Grained Image Difference Identification.
Selected Publications
You can find all of my articles on my Google Scholar profile.
(* = equal contribution)
🔍 Factuality and Faithfulness
MotionEdit: Benchmarking and Learning Motion-Centric Image Editing
Published in CVPR 2026, 2025
We introduce MotionEdit, a novel dataset and benchmark for motion-centric image editing. We also propose MotionNFT (Motion-guided Negative-aware FineTuning), a post-training framework with motion alignment rewards to guide models on motion editing task.
CompAlign: Improving Compositional Text-to-Image Generation with a Complex Benchmark and Fine-Grained Feedback
Published in Arxiv, 2025
Sequence-Level Certainty Reduces Hallucination In Knowledge-Grounded Dialogue Generation
Published in ICLR SeT LLM, 2023
🔍 Fairness
*InsideOut: Measuring and Mitigating Insider–Outsider Bias in Interview Script Generation
Published in ACL, 2026
*[Best Paper, NAACL TrustNLP 2024] White Men Lead, Black Women Help? Benchmarking and Mitigating Language Agency Social Biases in LLMs
Published in ACL, 2025
The Male CEO and the Female Assistant: Evaluation and Mitigation of Gender Biases in Text-To-Image Generation of Dual Subjects
Published in ACL, 2025
The Factuality Tax of Diversity-Intervened Text-to-Image Generation: Benchmark and Fact-Augmented Intervention
Published in EMNLP, 2024
Survey of Bias In Text-to-Image Generation: Definition, Evaluation, and Mitigation
Published in Arxiv, 2024
Are Personalized Stochastic Parrots More Dangerous? Evaluating Persona Biases in Dialogue Systems
Published in EMNLP, 2023
‘Kelly is a Warm Person, Joseph is a Role Model’: Gender Biases in LLM-Generated Reference Letters
Published in EMNLP, 2023
🔍 Safety and Privacy
Not Every Token Needs Forgetting: Selective Unlearning to Limit Change in Utility in Large Language Model Unlearning
Published in EMNLP, 2025
LUME: LLM Unlearning with Multitask Evaluations
Published in EMNLP, 2025