🧑‍🔬 Da Zheng

🏛️ Affiliations

🎓Hefei University of TechnologyHefei · China2023–2026 · 2 paperscurrent🔬Research Institute of Petroleum Exploration and DevelopmentBeijing · China2022–2023 · 4 papers🏢AmazonSeattle · United States2021–2022 · 3 papers🎓Southeast UniversityNanjing · China2006–2015 · 3 papers🎓Chengdu University of TechnologyChengdu · China2011–2024 · 2 papers🎓Johns Hopkins UniversityBaltimore · United States2014–2022 · 2 papers🔬State Key Laboratory of Digital Medical EngineeringNanjing · China2006–2007 · 2 papers

From the bylines of their papers (OpenAlex, 2026-10-14), ORCID and Wikidata; institutions on papers are matched by machine. 18 more institutions on single bylines are not listed.

🧭 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. LightThinker: Thinking Step-by-Step Compression (2025 · 50 citations)
   Slow, o1-style reasoning makes LLMs produce very long token sequences. Because attention cost grows quadratically and the KV Cache grows linearly with context, these outputs are expensive in memory and time. LightThinker addresses this by teaching a model to condense each intermediate reasoning step as it goes. After…
2. Untie the Knots: An Efficient Data Augmentation Strategy for Long-Context Pre-Training in Language Models (2025 · 7 citations)
   Extending large language models to very long contexts is difficult for three reasons. Natural documents of 32K–128K tokens are scarce. Upsampling long data shifts the training distribution and can hurt short-context skills. Attention over long sequences is also expensive. The authors propose Untie the Knots (UtK), an…
3. KnowRL: Exploring Knowledgeable Reinforcement Learning for Factuality (2026 · 5 citations)
   Slow-thinking LLMs such as DeepSeek-R1 distillations reason well but hallucinate heavily. One reason is that standard RL rewards only the final answer, so invented reasoning steps still get reinforced whenever the answer happens to be right. KnowRL addresses this by adding factual supervision to the reasoning trace…
4. What Makes AI Research Replicable? Executable Knowledge Graphs as Scientific Knowledge Representations (2026)
   LLM agents that try to reproduce AI papers often fail to produce working code. Papers leave out implementation specifics, and standard RAG does not recover technical details buried in cited works or the code signals in related repositories. The authors propose Executable Knowledge Graphs (xKG), a plug-in knowledge…

🤝 Co-authors here

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

• Huajun Chen ×3
• Ningyu Zhang ×3
• Lun Du ×2
• Shuofei Qiao ×2
• Yujie Luo ×2
• Yuqi Zhu ×2
• Debing Zhang
• Jintian Zhang
• Junfeng Tian
• Mengshu Sun
• Rui Wang
• Yang Chen

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 ×2, ACL 2025
• Conference on Empirical Methods in Natural Language Processing (EMNLP): EMNLP 2025

🔗 Cited by

Papers of our library cite their works 55 times (of the 149,439 papers whose bibliographies we read); 27 citing papers with a page here →.

Updated 2 November 2026

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