Xin HE's Website

Share papers, techniques, and wonderful life

RouteMark: 基于路由行为指纹的模型合并知识产权归属 | A Fingerprint for IP Attribution in Routing-based Model Merging

2 min read · March 12, 2026

2026 · Model-Merging MoE Intellectual-Property Fingerprint Security · research
AutoHete: 面向大语言模型的自动化高效异构训练系统 | An Automatic and Efficient Heterogeneous Training System for LLMs

2 min read · March 12, 2026

2026 · LLM Distributed-Training Heterogeneous-Computing Systems · research
GM-Skip: 基于度量引导的 Transformer 块跳过策略加速视觉语言模型 | Metric-Guided Transformer Block Skipping for Efficient VLMs

2 min read · March 12, 2026

2026 · VLM Transformer Efficiency Inference-Acceleration · research
Ghost in the Cloud: 地理分布式大模型训练的安全隐患 | Your Geo-distributed LLM Training is Easily Manipulated

2 min read · March 12, 2026

2026 · LLM Security Federated-Learning Distributed-Training ICLR · research
Lang-PINN: 从自然语言到物理信息神经网络的多智能体框架 | From Language to PINNs via a Multi-Agent Framework

2 min read · March 12, 2026

2026 · PINN LLM Multi-Agent Physics-Informed ICLR · research