Agora | Research Hub

Citations & Impact

All-time

Citations

938

H-index

11

i10-index

12

Publications

20

Co-authors

7

list available

Contact

Emaillilijun@pjlab.org.cn GitHubOpen ↗

Publications

8 items

SEARL: Joint Optimization of Policy and Tool Graph Memory for Self-Evolving Agents

2026

Cited

0

Seeing with You: Perception-Reasoning Coevolution for Multimodal Reasoning

2026

Cited

0

TreeTeaming: Autonomous Red-Teaming of Vision-Language Models via Hierarchical Strategy Exploration

2026

Cited

0

Stable Adaptive Thinking via Advantage Shaping and Length-Aware Gradient Regulation

2026

Cited

0

DeepSight: An All-in-One LM Safety Toolkit

2026

Cited

0

ADORA: Training Reasoning Models with Dynamic Advantage Estimation on Reinforcement Learning

2026

Cited

2

Toward Efficient Agents: Memory, Tool learning, and Planning

2026

Cited

1

ToolSafe: Enhancing Tool Invocation Safety of LLM-based agents via Proactive Step-level Guardrail and Feedback

2026

Cited

0

Resume (English only)

Academic Achievements

Paper 'PURE' accepted by NeurIPS 2025
Papers 'LARF', 'Visco-attack', and 'RMS' accepted by EMNLP 2025
Paper 'WorldSimBench' accepted by ICML 2025
Paper 'T2ISafety' accepted by CVPR 2025
Papers 'SALAD-Bench' and 'PsySafe' accepted by ACL 2024; 'PsySafe' awarded Outstanding Paper
Paper 'RenderIH' accepted by ICCV 2023
Ranked 1st in Action Detection track at EPIC@CVPR2022 Workshop (2022)
Ranked 2nd in FreiHAND Competition (2022)
Released an easy-to-use Python framework for generating adversarial jailbreak prompts
Co-first or corresponding author on multiple preprints and conference papers on AI safety, multimodal understanding/generation, and agent safety

Co-authors

7 total