Jiancan Wu

Quantile Advantage Estimation: Stabilizing RLVR for LLM Reasoning

Beyond Magnitude: Leveraging Direction of RLVR Updates for LLM Reasoning

PEA-DPO: Perception-Enhanced Alignment Direct Preference Optimization for MLLMs Alignment

Mitigating Reward Hacking in LLM-based Recommendation: A Preference Optimization Approach

bi-GRPO: Bidirectional Optimization for Jailbreak Backdoor Injection on LLMs

MLLMEraser: Achieving Test-Time Unlearning in Multimodal Large Language Models through Activation Steering

SINGER: Leveraging Semantic Identifier Hierarchies for Generative Recommendation

Attention Localization Through Separator Tokens: Unlocking Long Numerical Sequence Processing in LLMs

Bridging Perception and Reasoning: Token Reweighting for RLVR in Multimodal LLMs

Unified Parameter-Efficient Unlearning for LLMs

Think before Recommendation: Autonomous Reasoning-enhanced Recommender

Uncertainty-aware Guided Diffusion for Missing Data in Sequential Recommendation

Fading to Grow: Growing Preference Ratios via Preference Fading Discrete Diffusion for Recommendation

RePO: Understanding Preference Learning Through ReLU-Based Optimization

Towards Robust Alignment of Language Models: Distributionally Robustifying Direct Preference Optimization

On Efficiency-Effectiveness Trade-off of Diffusion-based Recommenders

Larger or Smaller Reward Margins to Select Preferences for LLM Alignment?

$\alpha$-DPO: Adaptive Reward Margin is What Direct Preference Optimization Needs

AlphaDPO: Adaptive Reward Margin for Direct Preference Optimization