Built with and by Teycir Ben Soltane•
How to Use•FAQ•GitHub•arXiv.org•
Share:
ArXivCSExplorer
☆☆Bookmarks🏆RSSHow to UseFAQ
Home/Authors/Pan Li

Pan Li

6 indexed papers

Recent (6 mo)
3
With code
0
Influential cites
0
Benchmarked
0

Publications per year

6
26

Top categories

AI×4NLP×3Crypto×3Distributed×1ML×1Software Eng.×1

Frequent co-authors

Haoyu Wang3×
Mufei Li2×
Shikun Liu2×
Dongqi Fu2×
Yinglong Xia2×
Hong Li2×

Research Timeline

2026
Knowdit: Agentic Smart Contract Vulnerability Detection with Auditing Knowledge Summarization

Knowdit is a knowledge-driven, agentic framework that significantly improves smart contract vulnerability detection by modeling shared DeFi semantics and leveraging historical audit knowledge.

How Far Are VLMs from Privacy Awareness in the Physical World? An Empirical Study

The paper introduces ImmersedPrivacy, an interactive audio-visual framework, and finds that current Vision-Language Models (VLMs) deployed in physical environments suffer from significant deficits in privacy awareness, particularly in complex or socially ambiguous settings.

One Turn Too Late: Response-Aware Defense Against Hidden Malicious Intent in Multi-Turn Dialogue

The paper introduces TurnGate, a response-aware defense mechanism that detects the earliest turn in a multi-turn dialogue where the accumulated interaction enables a harmful action, significantly improving malicious intent detection.

Towards Direct Latent-Space Synthesis for Parallel Branches in LLM-Agent Workflows

Introduce Parallel-Synthesis, a framework enabling a synthesizer to directly consume parallel agent branches' KV caches, improving efficiency and performance.

KVEraser: Learning to Steer KV Cache for Efficient Localized Context Erasing

The paper presents KVEraser, a learned KV-cache editing method for efficient localized context erasing in long-context LLM applications.

HCMS: Head-Chunked Multi-Stream Pipeline for Communication-Computation Overlap in Long-Sequence Parallel Attention

This paper proposes Head-Chunked Multi-Stream Pipeline (HCMS) to exploit the computational independence of multi-head attention and achieve fine-grained communication-computation overlap, resulting in speedups of up to 17.5% over the Ulysses baseline.

Highlighted terms show continued research focus across papers

Papers

cs.DCEmpiricalRecentJul 2, 2026

HCMS: Head-Chunked Multi-Stream Pipeline for Communication-Computation Overlap in Long-Sequence Parallel Attention

Chao Yuan, Pan Li, Yingnan Sun, Jing Liu

This paper proposes Head-Chunked Multi-Stream Pipeline (HCMS) to exploit the computational independence of multi-head attention and achieve fine-grained communication-computation overlap, resulting in…

View →
cs.CLcs.LGEmpirical
Recent
Jun 15, 2026

KVEraser: Learning to Steer KV Cache for Efficient Localized Context Erasing

Mufei Li, Shikun Liu, Dongqi Fu, Haoyu Wang +4 more

The paper presents KVEraser, a learned KV-cache editing method for efficient localized context erasing in long-context LLM applications.

View →
cs.AIcs.CLEmpiricalRecentJun 12, 2026

Towards Direct Latent-Space Synthesis for Parallel Branches in LLM-Agent Workflows

Shikun Liu, Mufei Li, Dongqi Fu, Haoyu Wang +4 more

Introduce Parallel-Synthesis, a framework enabling a synthesizer to directly consume parallel agent branches' KV caches, improving efficiency and performance.

View →
cs.CLcs.AIcs.CRMay 7, 2026

One Turn Too Late: Response-Aware Defense Against Hidden Malicious Intent in Multi-Turn Dialogue

Xinjie Shen, Rongzhe Wei, Peizhi Niu, Haoyu Wang +5 more

The paper introduces TurnGate, a response-aware defense mechanism that detects the earliest turn in a multi-turn dialogue where the accumulated interaction enables a harmful action, significantly impr…

View →
cs.CRcs.AIMay 6, 2026

How Far Are VLMs from Privacy Awareness in the Physical World? An Empirical Study

Junran Wang, Xinjie Shen, Zehao Jin, Pan Li

The paper introduces ImmersedPrivacy, an interactive audio-visual framework, and finds that current Vision-Language Models (VLMs) deployed in physical environments suffer from significant deficits in…

View →
cs.CRcs.AIcs.SEMar 27, 2026

Knowdit: Agentic Smart Contract Vulnerability Detection with Auditing Knowledge Summarization

Ziqiao Kong, Wanxu Xia, Chong Wang, Yi Lu +4 more

Knowdit is a knowledge-driven, agentic framework that significantly improves smart contract vulnerability detection by modeling shared DeFi semantics and leveraging historical audit knowledge.

View →