🧑‍🔬 Juntao Li

🏛️ Affiliations

🎓Soochow UniversitySuzhou · China2004–2026 · 80 papersEducation 2015–2020 · Ph.d, Peking Universitycurrent🎓Beijing Institute of TechnologyBeijing · China2026 · 1 papercurrent🎓Peking UniversityBeijing · China2007–2021 · 24 papers🎓Guizhou UniversityGuiyang · China2009–2010 · 4 papers🎓Dalian University of TechnologyDalian · China2025 · 3 papers🏛️Chinese Academy of SciencesBeijing · China2024–2025 · 2 papers🎓Harbin Institute of TechnologyHarbin · China2024–2025 · 2 papers🔬Laboratoire de Chimie MoléculaireStrasbourg · France2025 · 2 papers🏥First Affiliated Hospital of Soochow UniversitySuzhou · China2024 · 2 papers🎓National University of SingaporeSingapore · Singapore2020–2022 · 2 papers

From the bylines of their papers (OpenAlex, 2026-10-14), ORCID and Wikidata; institutions on papers are matched by machine. 10 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. OPT-Tree: Speculative Decoding with Adaptive Draft Tree Structure (2025 · 32 citations)
   Speculative decoding speeds up LLM inference without changing outputs: a cheap draft model proposes several tokens, and the large target model checks them all in one parallel pass.
2. Timo: Towards Better Temporal Reasoning for Language Models (2024 · 19 citations)
   Language models still reason poorly about time. Earlier fixes targeted narrow settings such as time-sensitive question answering, so they carry over badly to other temporal problems. This work looks for one framework that covers many kinds of temporal reasoning. The authors examine 38 subtasks from the Wang & Zhao…
3. Can Diffusion Model Achieve Better Performance in Text Generation ? Bridging the Gap between Training and Inference ! (2023 · 18 citations)
   Diffusion models for text embed discrete tokens into a continuous space and then denoise them. Their training and inference behave differently, and this hurts output quality. During training, each noisy state is computed directly from the clean data. At inference, the model must build on its own earlier predictions…
4. Efficient Reasoning for LLMs through Speculative Chain-of-Thought (2026 · 6 citations)
   Reasoning LLMs such as Deepseek-R1 are slow because large models produce very long chains of thought (CoT).
5. IS-CoT: Breaking the Long-form Generation Collapse via Interleaved Structural Thinking (2026 · 1 citation)
   The paper tackles a weakness of LLMs in open-ended long-form writing: once the requested length goes beyond roughly 2,000 words, both quality and adherence to the target length drop sharply.
6. When Is Thinking Enough? Early Exit via Sufficiency Assessment for Efficient Reasoning (2026 · 1 citation)
   Large reasoning models tend to overthink: they keep re-verifying or trying other strategies after they already have a correct solution, which wastes tokens.
7. Beyond Task-Level Context: Class-Conditional Context Vectors for Implicit In-Context Learning (2026)
   Implicit in-context learning replaces prompt demonstrations with a context vector added to a model's internal activations.
8. Crossing the Reward Bridge: Expanding Reinforcement Learning with Verifiable Rewards Across Diverse Domains (2026)
   RLVR works well for math and code, where a simple rule can check whether an answer is right. Such checkers fail on free-form answers, and the authors show that rule-based rewards can even get worse as more mixed-domain training data is added. The paper replaces the rule with a generative verifier. An LLM compares the…
9. Data Foundations of Long-Context Language Models: A Survey (2026)
   Most earlier surveys of long-context language models (LCMs, here meaning models with windows above 128K tokens) focus on architecture and efficient training, so the role of data has been largely overlooked.
10. When to Trust Tools? Adaptive Tool Trust Calibration For Tool-Integrated Math Reasoning (2026)
    Tool-Integrated Reasoning (TIR) models solve math problems by interleaving reasoning with code execution. The authors show that these models often handle disagreements between their own reasoning and the tool's output poorly. They audited more than 32k trajectories with an LLM and found that most conflicts are…

All works

11 works of theirs in our library, the most cited first; 10 with a page here, the others link their original until we summarise them.

1. Generating Classical Chinese Poems via Conditional Variational Autoencoder and Adversarial Training ↗Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing · no summary yetJuntao Li et al.2018 · 86 citations
2. OPT-Tree: Speculative Decoding with Adaptive Draft Tree StructureTransactions of the Association for Computational Linguistics (TACL)Jikai Wang, Yi Su, Juntao Li, et al.2025 · 32 citations
3. Timo: Towards Better Temporal Reasoning for Language ModelsarXiv:2406.14192Zhaochen Su, Jun Zhang, Tong Zhu, et al.2024 · 19 citations
4. Can Diffusion Model Achieve Better Performance in Text Generation ? Bridging the Gap between Training and Inference !Findings of the Association for Computational LinguisticsZecheng Tang, Pinzheng Wang, Keyan Zhou, et al.2023 · 18 citations
5. Efficient Reasoning for LLMs through Speculative Chain-of-ThoughtFindings of the Association for Computational LinguisticsJikai Wang, Juntao Li, Jianye Hou, et al.2026 · 6 citations
6. IS-CoT: Breaking the Long-form Generation Collapse via Interleaved Structural ThinkingAnnual Meeting of the Association for Computational Linguistics (ACL)Zechen Sun, Yuyang Sun, Zecheng Tang, et al.2026 · 1 citation
7. When Is Thinking Enough? Early Exit via Sufficiency Assessment for Efficient ReasoningAnnual Meeting of the Association for Computational Linguistics (ACL)Yang Xiang, Yixin Ji, Ruotao Xu, et al.2026 · 1 citation
8. Beyond Task-Level Context: Class-Conditional Context Vectors for Implicit In-Context LearningFindings of the Association for Computational LinguisticsJianxin Zhang, Yilu Hu, Teng Liu, et al.2026
9. Crossing the Reward Bridge: Expanding Reinforcement Learning with Verifiable Rewards Across Diverse DomainsAnnual Meeting of the Association for Computational Linguistics (ACL)Yi Su, Dian Yu, Linfeng Song, et al.2026
10. Data Foundations of Long-Context Language Models: A SurveyTransactions of the Association for Computational Linguistics (TACL)Zechen Sun, Yuyang Sun, Zhaochen Su, et al.2026
11. When to Trust Tools? Adaptive Tool Trust Calibration For Tool-Integrated Math ReasoningFindings of the Association for Computational LinguisticsRuotao Xu, Yixin Ji, Yu Luo, et al.2026

🤝 Co-authors here

12 of 33 co-authors with a page, the most papers together first.

• Min Zhang ×9
• Zecheng Tang ×3
• Jikai Wang ×2
• Ruotao Xu ×2
• Wenliang Chen ×2
• Yi Su ×2
• Yixin Ji ×2
• Yuyang Sun ×2
• Zechen Sun ×2
• Zhaochen Su ×2
• Ao Zhou
• Dian Yu

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) ×4: ACL 2026 ×3
• Findings of the Association for Computational Linguistics ×4: Findings of the Association for Computational Linguistics 2026 ×3, Findings of the Association for Computational Linguistics 2023
• Transactions of the Association for Computational Linguistics (TACL) ×2: Transactions of the Association for Computational Linguistics (TACL) 2026, Transactions of the Association for Computational Linguistics (TACL) 2025
• Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing

🔗 Cited by

Papers of our library cite their works 80 times (of the 149,439 papers whose bibliographies we read); 16 citing papers with a page here →.

Updated 2 November 2026

↑ back to top

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