Yang Shi

AVoCaDO: An Audiovisual Video Captioner Driven by Temporal Orchestration

The Unseen Bias: How Norm Discrepancy in Pre-Norm MLLMs Leads to Visual Information Loss

BaseReward: A Strong Baseline for Multimodal Reward Model

When Modalities Conflict: How Unimodal Reasoning Uncertainty Governs Preference Dynamics in MLLMs

MME-Unify: A Comprehensive Benchmark for Unified Multimodal Understanding and Generation Models

VidBridge-R1: Bridging QA and Captioning for RL-based Video Understanding Models with Intermediate Proxy Tasks

OpenGPT-4o-Image: A Comprehensive Dataset for Advanced Image Generation and Editing

Beyond Rational Illusion: Behaviorally Realistic Strategic Classification

Transformers with Endogenous In-Context Learning: Bias Characterization and Mitigation

RealUnify: Do Unified Models Truly Benefit from Unification? A Comprehensive Benchmark

Identifying Outcome-Oriented Root Causes via Cross Regression

A Unified and Data-Efficient Framework for Out-of-Distribution and Generalization

Hybrid Attribution Priors for Explainable and Robust Model Training

MM-RLHF: The Next Step Forward in Multimodal LLM Alignment