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

Lijun Li

8 indexed papers

Recent (6 mo)
8
With code
0
Influential cites
0
Benchmarked
0

Publications per year

8
26

Top categories

AI×7NLP×3Multiagent×2Distributed×1Crypto×1

Frequent co-authors

Pengyu Zhu4×
Jing Shao4×
Hao Li2×
Jingkun An2×
Zijun Song2×
Rui Li2×

Research Timeline

2026
A Unified Framework for the Evaluation of LLM Agentic Capabilities

The paper introduces a unified framework to fairly evaluate LLM agentic capabilities by standardizing diverse benchmarks and separating the effects of the LLM model from the surrounding framework and environment.

A Unified Framework for the Evaluation of LLM Agentic Capabilities

The paper introduces a unified framework to fairly evaluate LLM agentic capabilities by standardizing diverse benchmarks and separating the effects of the LLM model from the surrounding framework and environment.

AgentSchool: An LLM-Powered Multi-Agent Simulation for Education

The paper introduces AgentSchool, an advanced LLM-powered multi-agent simulator that models learning as state transitions to provide a robust, ethically viable testbed for educational research and pedagogical reform.

AgentSchool: An LLM-Powered Multi-Agent Simulation for Education

The paper introduces AgentSchool, an advanced LLM-powered multi-agent simulator that models learning as state transitions to provide a robust, ethically viable testbed for educational research and pedagogical reform.

SafeSteer: Localized On-Policy Distillation for Efficient Safety Alignment

SafeSteer proposes a localized on-policy distillation method that restricts safety alignment to specific safety tokens, thereby achieving strong safety performance with minimal degradation to general capabilities and significantly reducing data requirements.

SafeSteer: Localized On-Policy Distillation for Efficient Safety Alignment

SafeSteer proposes a localized on-policy distillation method that restricts safety alignment to specific safety tokens, thereby achieving strong safety performance with minimal degradation to general capabilities and significantly reducing data requirements.

JANUS: Foreseeing Latent Risk for Long-Horizon Agent Safety

The paper proposes Janus, a framework for long-horizon agent safety using multi-agent simulation and CoAA-RL, which trains guards to anticipate risks and make safety judgments based on both observed prefix and anticipated future.

StrataCL: Fabric-Native Communication Library for Production Supernodes

StrataCL is a zero-redundancy communication library for production supernodes that introduces registration-on-allocation for user-buffer direct communication and designs communication operators for workload-balanced NPU-core partitioning and NPU-driven SDMA offloading, improving collective and dispatch/combine bus bandwidth, as well as inference throughput, P99 TTFT, and training iteration time on Huawei CloudMatrix384 and three production workloads.

Highlighted terms show continued research focus across papers

Papers

cs.DCEmpiricalRecentJul 29, 2026

StrataCL: Fabric-Native Communication Library for Production Supernodes

Tiancheng Hu, Jin Qin, Yuzheng Wang, Ke Liu +13 more

StrataCL is a zero-redundancy communication library for production supernodes that introduces registration-on-allocation for user-buffer direct communication and designs communication operators for wo…

View →
cs.AIcs.CLcs.CREmpirical
Recent
Jul 22, 2026

JANUS: Foreseeing Latent Risk for Long-Horizon Agent Safety

Yuan Xiong, Linji Hao, Shizhu He, Yequan Wang +1 more

The paper proposes Janus, a framework for long-horizon agent safety using multi-agent simulation and CoAA-RL, which trains guards to anticipate risks and make safety judgments based on both observed p…

View →
cs.AIcs.CLRecentJun 1, 2026

SafeSteer: Localized On-Policy Distillation for Efficient Safety Alignment

Hao Li, Jingkun An, Zijun Song, Pengyu Zhu +7 more

SafeSteer proposes a localized on-policy distillation method that restricts safety alignment to specific safety tokens, thereby achieving strong safety performance with minimal degradation to general…

View →
cs.AIcs.CLRecentJun 1, 2026

SafeSteer: Localized On-Policy Distillation for Efficient Safety Alignment

Hao Li, Jingkun An, Zijun Song, Pengyu Zhu +7 more

SafeSteer proposes a localized on-policy distillation method that restricts safety alignment to specific safety tokens, thereby achieving strong safety performance with minimal degradation to general…

View →
cs.AIcs.MARecentMay 28, 2026

AgentSchool: An LLM-Powered Multi-Agent Simulation for Education

Yulei Ye, Wenhao Li, Zhong Wen, Yunshu Huang +22 more

The paper introduces AgentSchool, an advanced LLM-powered multi-agent simulator that models learning as state transitions to provide a robust, ethically viable testbed for educational research and ped…

View →
cs.AIcs.MARecentMay 28, 2026

AgentSchool: An LLM-Powered Multi-Agent Simulation for Education

Yulei Ye, Wenhao Li, Zhong Wen, Yunshu Huang +22 more

The paper introduces AgentSchool, an advanced LLM-powered multi-agent simulator that models learning as state transitions to provide a robust, ethically viable testbed for educational research and ped…

View →
cs.AIRecentMay 27, 2026

A Unified Framework for the Evaluation of LLM Agentic Capabilities

Pengyu Zhu, Lijun Li, Yaxing Lyu, Qianxin Luo +7 more

The paper introduces a unified framework to fairly evaluate LLM agentic capabilities by standardizing diverse benchmarks and separating the effects of the LLM model from the surrounding framework and…

View →
cs.AIRecentMay 27, 2026

A Unified Framework for the Evaluation of LLM Agentic Capabilities

Pengyu Zhu, Lijun Li, Yaxing Lyu, Qianxin Luo +7 more

The paper introduces a unified framework to fairly evaluate LLM agentic capabilities by standardizing diverse benchmarks and separating the effects of the LLM model from the surrounding framework and…

View →