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…
Can We Predict Before Executing Machine Learning Agents? (2026 · 2 citations)
ML agents usually improve their code through repeated generate-execute-feedback cycles. Every candidate has to be trained and run, so a single MLE-Bench run can take many hours. The authors ask whether an LLM can act as an implicit world model and judge which solution will perform better before any code runs. They…
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…

All works

4 works of theirs in our library, the most cited first; 3 with a page here, the others link their original until we summarise them.

LightThinker: Thinking Step-by-Step CompressionConference on Empirical Methods in Natural Language Processing (EMNLP)Jintian Zhang, Yuqi Zhu, Mengshu Sun, et al.2025 · 50 citations
Can We Predict Before Executing Machine Learning Agents?Annual Meeting of the Association for Computational Linguistics (ACL)Jingsheng Zheng, Jintian Zhang, Yujie Luo, et al.2026 · 2 citations
What Makes AI Research Replicable? Executable Knowledge Graphs as Scientific Knowledge RepresentationsAnnual Meeting of the Association for Computational Linguistics (ACL)Yujie Luo, Zhuoyun Yu, Xuehai Wang, et al.2026
StructMem: Structured Memory for Long-Horizon Behavior in LLMs ↗Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers) · no summary yetBuqiang Xu et al.2026

Co-authors here

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

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

Papers of our library cite their works 47 times (of the 149,439 papers whose bibliographies we read); 19 citing papers with a page here →.

Updated 1 November 2026