Jie Wang

15 papers Southeast University

• Accurate KV Cache Eviction via Anchor Direction Projection for Efficient LLM Inference
• ArchCAD-400K: A Large-Scale CAD drawings Dataset and New Baseline for Panoptic Symbol Spotting
• AttentionPredictor: Temporal Patterns Matter for KV Cache Compression
• Benchmarking End-To-End Performance of AI-Based Chip Placement Algorithms
• Can Class-Priors Help Single-Positive Multi-Label Learning?
• Dynamic Configuration for Cutting Plane Separators via Reinforcement Learning on Incremental Graph
• High-Performance Arithmetic Circuit Optimization via Differentiable Architecture Search
• LogicTree: Improving Complex Reasoning of LLMs via Instantiated Multi-step Synthetic Logical Data
• MeshCoder: LLM-Powered Structured Mesh Code Generation from Point Clouds
• Mixture-of-Experts Operator Transformer for Large-Scale PDE Pre-Training
• OptiTree: Hierarchical Thoughts Generation with Tree Search for LLM Optimization Modeling
• Real-World Reinforcement Learning of Active Perception Behaviors
• RoME: Domain-Robust Mixture-of-Experts for MILP Solution Prediction across Domains
• STNet: Spectral Transformation Network for Solving Operator Eigenvalue Problem
• SymMaP: Improving Computational Efficiency in Linear Solvers through Symbolic Preconditioning