Mingkui Tan

ZeroSiam: An Efficient Asymmetry for Test-Time Entropy Optimization without Collapse

R-Stitch: Dynamic Trajectory Stitching for Efficient Reasoning

Beyond Fast and Slow: Cognitive-Inspired Elastic Reasoning for Large Language Models

Beyond Model Scaling: Test-Time Intervention for Efficient Deep Reasoning

Layer Collaborative Low-Rank Decomposition with Automatic Rank Search for LLM Compression

One Stone Three Birds: Training-free Core-context-aware Attention for Efficient LLM Prefilling, Decoding, and KV Caching

Next-Token Gradient Sensitivity Probing for LLM Hallucination Detection

AIGCoder 1.0: Locally-Enhanced Language Modeling with Explicit and Structured Knowledge Memory

Decoupling of Experts: A Knowledge-Driven Architecture for Efficient LLMs

Physics-Driven Spatiotemporal Modeling for AI-Generated Video Detection

Open-World Drone Active Tracking with Goal-Centered Rewards

Curse of High Dimensionality Issue in Transformer for Long Context Modeling

Deep Electromagnetic Structure Design Under Limited Evaluation Budgets

Core Context Aware Transformers for Long Context Language Modeling

Test-time Adapted Reinforcement Learning with Action Entropy Regularization

Continual Knowledge Adaptation for Reinforcement Learning

Enhancing Perception Capabilities of Multimodal LLMs with Training-Free Fusions

Test-Time Learning for Large Language Models

Core Context Aware Attention for Long Context Language Modeling

Deep Progressive Search for Electromagnetic Structure Design Under Limited Evaluation Budgets