Wenbo Su

Tricks or Traps? A Deep Dive into RL for LLM Reasoning

Expert Divergence Learning for MoE-based Language Models

COMI: Coarse-to-fine Context Compression via Marginal Information Gain

DESIGNER: Design-Logic-Guided Multidisciplinary Data Synthesis for LLM Reasoning

Asymmetric Proximal Policy Optimization: mini-critics boost LLM reasoning

MeSH: Memory-as-State-Highways for Recursive Transformers

RAVR: Reference-Answer-guided Variational Reasoning for Large Language Models

Reconstructing KV Caches with Cross-Layer Fusion for Enhanced Transformers

Beyond Safe Answers: A Benchmark for Evaluating True Risk Awareness in Large Reasoning Models

USB: A Comprehensive and Unified Safety Evaluation Benchmark for Multimodal Large Language Models

One Sample to Rule Them All: Extreme Data Efficiency in RL Scaling

Towards Visualization-of-Thought Jailbreak Attack against Large Visual Language Models

MTU-Bench: A Multi-granularity Tool-Use Benchmark for Large Language Models

M2rc-Eval: Massively Multilingual Repository-level Code Completion Evaluation

Truly Safe & Truly Helpful: Achieving Harmonious Balance for Large Language Model

Adaptive dense reward:Understanding the Gap Between Action and Reward Space in Alignment