🧑‍🔬 Zhengqi Wen

🏛️ Affiliations

🎓Tsinghua UniversityBeijing · China2026 · 19 paperscurrent🔬Beijing National Research Center for Information Science and TechnologyBeijing · China2026 · 5 paperscurrent🏛️Chinese Academy of SciencesBeijing · China2026 · 2 paperscurrent🔬Institute of AutomationBeijing · China2026 · 2 paperscurrent🎓University of Chinese Academy of SciencesBeijing · China2026 · 1 papercurrent

From the bylines of their papers (OpenAlex, 2026-10-14), ORCID and Wikidata; institutions on papers are matched by machine.

🧭 Fields, topics and concepts

Of its works in our library, with the number of works on each.

📅 Works by year

2020s

📄 Papers

The most cited first (OpenAlex and our library).

1. SPARK: Strategic Policy-Aware Exploration via Dynamic Branching for Long-Horizon Agentic Learning (2026 · 11 citations)
   Training LLM agents with reinforcement learning on long, multi-step tasks is difficult because successful trajectories are rare, especially when compute is limited.
2. ATLAS: Orchestrating Heterogeneous Models and Tools for Multi-Domain Complex Reasoning (2026 · 8 citations)
   As both LLMs and external tools multiply, choosing which model should handle a query and with which tool becomes a large combinatorial search problem.
3. TemplateRL: Structured Template-Guided Reinforcement Learning for LLM Reasoning (2026 · 8 citations)
   RL methods such as GRPO strengthen LLM reasoning, but they learn only from rollouts the model samples freely from its own distribution.
4. Beyond Examples: Towards Automated Thought-level In-Context Reasoning for Large Language Models (2026)
   Standard in-context learning (ICL) teaches LLMs by showing worked examples. The model then copies their surface form, which breaks down on hard multi-step problems. Results also depend heavily on how the demonstrations are chosen and ordered. ThoughtICR swaps concrete examples for abstract reasoning strategies.
5. CAIR: Causal Adaptive Information-based Reinforcement Learning for Multimodal Emotion Reasoning (2026)
   The paper targets a weakness of multimodal emotion reasoning models that the authors call causal degeneracy.
6. Calibration-Aware Policy Optimization for Reasoning LLMs (2026)
   RL fine-tuning with verifiable rewards, especially GRPO, makes reasoning LLMs more accurate but also overconfident.

🤝 Co-authors here

12 of 14 co-authors with a page, the most papers together first.

• Jianhua Tao ×5
• Jinyang Wu ×4
• Shuai Zhang ×4
• Chonghua Liao ×2
• Haoran Luo ×2
• Ling Yang ×2
• Mingkuan Feng ×2
• Yuhao Shen ×2
• Bin Liu
• Feihu Che
• Shuo Yang
• Zheng Lian

Venues

Where their works here were published, the most works first, with the editions of those that have a page.

• Annual Meeting of the Association for Computational Linguistics (ACL) ×3: ACL 2026 ×3
• Findings of the Association for Computational Linguistics ×3: Findings of the Association for Computational Linguistics 2026 ×3

🔗 Cited by

Papers of our library cite their works 26 times (of the 149,439 papers whose bibliographies we read); 13 citing papers with a page here →.

Updated 2 November 2026

This page in Markdown → https://md.research.elamur.ai/free/en/authors/zhengqi-wen/