Jian Cheng

Ban&Pick: Ehancing Performance and Efficiency of MoE-LLMs via Smarter Routing

Open Set Opponent Modeling

DRef: A Benchmark with Diverse Referring Expressions for Object Comprehension of Vision-Language Models

Achieving Ultra-Low Latency and Lossless ANN-SNN Conversion through Optimal Elimination of Unevenness Error

On the Suboptimality of Semi-Markov Decision Process in Hierarchical Reinforcement Learning

Block Rotation is All You Need for MXFP4 Quantization

Mixed-Timestep Spiking Neural Networks with Temporal Alignment for Ultra-Low Latency Conversion

LLM-based Symbolic Regression with Tool-Augmented Multi-Objective Optimization

Bi-Level Knowledge Transfer for Multi-Task Multi-Agent Reinforcement Learning

Goal-Oriented Skill Abstraction for Offline Multi-Task Reinforcement Learning

DartQuant: Efficient Rotational Distribution Calibration for LLM Quantization

QAC:Quantization-Aware Conversion for Mixed-Timestep Spiking Neural Networks

Maximum Entropy Reinforcement Learning with Diffusion Policy

Evo-Step: Evolutionary Generation and Stepwise Validation for Optimizing LLMs in OR

Offline Opponent Modeling with Truncated Q-driven Instant Policy Refinement

Q-Mamba: Towards more efficient Mamba models via Post-Training Quantization

Ultra-Low Accumulation Precision Inference with Block Floating Point Arithmetic