5 papers
Autonomy-of-Heads: Data-Free Sparse Attention from Frozen Query-Key Geometry
Yehan Yang, Junyuan Shang, Yang Li +3
Long-context LLM inference is bottlenecked by quadratic attention computation and growing KV-cache costs. Existing sparse attention and KV-compression methods typically decide whic…
Deconstructing Off-Policy Ratios: Entropy-Normalized Trust Regions for Asynchronous Reinforcement Learning
Guanqun Zhao, Zijun Xie, Binbin Zheng +5
Asynchronous reinforcement learning (RL) accelerates large language model (LLM) post-training by overlapping rollout generation with policy optimization, but the resulting stale, o…
Beyond the Final Actor: Modeling the Dual Roles of Creator and Editor for Fine-Grained LLM-Generated Text Detection
Yang Li, Qiang Sheng, Zhengjia Wang +3
The misuse of large language models (LLMs) requires precise detection of synthetic text. Existing works mainly follow binary or ternary classification settings, which can only dist…
PhantomHunter: Detecting Unseen Privately-Tuned LLM-Generated Text via Family-Aware Learning
Yuhui Shi, Yehan Yang, Qiang Sheng +4
With the popularity of large language models (LLMs), undesirable societal problems like misinformation production and academic misconduct have been more severe, making LLM-generate…
From Judgment to Interference: Early Stopping LLM Harmful Outputs via Streaming Content Monitoring
Yang Li, Qiang Sheng, Yehan Yang +2
Though safety alignment has been applied to most large language models (LLMs), LLM service providers generally deploy a subsequent moderation as the external safety guardrail in re…