Homepage
AI Engineering
Paper Notes
Course
Interview
Blog
Paper Notes
论文阅读笔记 · 16 篇
agent_safety
DRIFT-Dynamic Rule-Based Defense with InjectionIsolation for Securing LLM Agents
2026.11.13
PromptArmor - Simple yet Effective Prompt Injection Defenses.pdf
2026.11.13
Security–Fidelity Tradeoffs-The Hidden Cost of Prompt Injection Defense
2026.11.13
ReSI-Recursive Safety Improvementtoward Resistant and Resilient AI
2026.11.12
DOVER- INTERVENTION-DRIVEN AUTO DEBUGGINGFOR LLM MULTI-AGENT SYSTEMS
2026.10.05
Infra
PREBLE_EFFICIENT DISTRIBUTED PROMPT SCHEDULING FOR LLM SERVING
2026.11.10
METIS_Fast Quality-Aware RAG Systems withConfiguration Adaptation
2026.11.05
TRIINFER_HYBRID DISAGGREGATED SCHEDULING FOR MULTIMODALLARGE LANGUAGE MODEL SERVING
2026.11.02
OSWORLD-HUMAN_BENCHMARKING THE EFFICIENCY OF COMPUTER-USEAGENTS
2026.10.28
TClone_Low-Latency Forking of Live GUI Environments for Computer-Use Agents
2026.10.20
PowerInfer_ Fast Large Language Model Serving with aConsumer-grade GPU
2026.10.08
THUNDERKITTENS_Simple, Fast, and Adorable Kernels
2026.10.05
Efficient Memory Management for Large Language Model Serving with PagedAttention
2026.10.01
MemGPT_Towards LLMs as Operating Systems
2026.10.01
Reasoning
Reasoning Pathologies in Large Language Models_A Diagnostic Perspective
2026.11.04
training
HybridFlow_A Flexible and Efficient RLHF Framework
2026.11.12