🧑‍🔬 Wentao Zhang

🏛️ Affiliations

🎓University of Electronic Science and Technology of ChinaChengdu · China2024–2027 · 8 papersEducation 2024–2027 · Master of Science in Computer Science and Technology, School of Computer Science and Engineering(School of Cyber Security))current🎓Peking UniversityBeijing · China2026 · 8 paperscurrent🏢ByteDanceBeijing · China2026 · 1 papercurrent🎓ShanghaiTech UniversityShanghai · China2026 · 1 papercurrent🎓Guangdong University of TechnologyGuangzhou · China2020–2024Education 2020–2024 · B.S. in Artificial Intelligence, School of Computer Science and Technology

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. Retrieval-Augmented Generation for AI-Generated Content: A Survey (2024 · 301 citations)
   Generative models for text, code, images, video and other media still have trouble with outdated knowledge, rare (long-tail) facts, leakage of private training data, and costly training and inference.
2. TC–RAG: Turing–Complete RAG's Case study on Medical LLM Systems (2025 · 25 citations)
   Retrieval-augmented generation can reduce hallucinations and stale knowledge in medical LLMs. However, existing RAG pipelines lack an explicit state variable that tells the system when to stop retrieving, so they can loop without converging or pile up noisy context. The authors propose TC–RAG, which they prove to be…
3. MM-Verify: Enhancing Multimodal Reasoning with Chain-of-Thought Verification (2025 · 19 citations)
   The paper targets two gaps in multimodal math reasoning. There is no strong verifier for checking answers from multimodal LLMs, and there is little long chain-of-thought (CoT) training data with images. The authors build two 7B models. MM-Verifier is an outcome reward model that judges whether a solution to an…
4. AgenticRAGTracer: A Hop-Aware Benchmark for Diagnosing Multi-Step Retrieval Reasoning in Agentic RAG (2026 · 1 citation)
   Popular multi-hop QA benchmarks such as HotpotQA, 2WikiMultihopQA and MuSiQue give only a question and its final answer.
5. CodeFlowBench: A Multi-turn, Iterative Benchmark for Complex Code Generation (2026 · 1 citation)
   Most code-generation benchmarks test LLMs in a single turn, but real software is built step by step from modular functions that reuse earlier ones. The authors call this incremental, reuse-driven process "codeflow" and present CodeFlowBench as the first benchmark for evaluating it.
6. SciFlow-Bench: Evaluating Structure-Aware Scientific Diagram Generation via Inverse Parsing (2026 · 1 citation)
   Text-to-image systems can produce scientific figures that look polished but are logically wrong: they skip steps, invent labels, or point arrows in the wrong direction.
7. Reinforcement Learning with Semantic Rewards Enables Low-Resource Language Expansion without Alignment Tax (2026)
   When LLMs are adapted to a poorly supported language, they often lose ground on skills they already had. This is known as the alignment tax. The authors blame supervised fine-tuning (SFT): its token-level likelihood objective forces the model to copy small, skewed datasets, which overwrites pretrained knowledge. They…
8. Rethinking Text-to-SQL: Dynamic Multi-turn SQL Interaction for Real-world Database Exploration (2026)
   Most Text-to-SQL benchmarks test one-shot translation of a fixed question into a read-only query. Real database work is different: users refine goals over several turns and often need to change data, not just read it. The authors introduce DySQL-Bench to evaluate LLM agents in this setting. It contains 1,072…
9. Training Verifier to Assessing Complex Real-World Tool-Use Trajectories (2026)
   LLM agents that handle multi-turn customer-service style tasks need training dialogues that reflect real user–agent collaboration.

All works

11 works of theirs in our library, the most cited first; 9 with a page here, the others link their original until we summarise them.

1. Retrieval-Augmented Generation for AI-Generated Content: A SurveyarXiv:2402.19473Penghao Zhao, Hailin Zhang, Qinhan Yu, et al.2024 · 301 citations
2. TC–RAG: Turing–Complete RAG's Case study on Medical LLM SystemsAnnual Meeting of the Association for Computational Linguistics (ACL)Xinke Jiang, Yue Fang, Rihong Qiu, et al.2025 · 25 citations
3. MM-Verify: Enhancing Multimodal Reasoning with Chain-of-Thought VerificationAnnual Meeting of the Association for Computational Linguistics (ACL)Linzhuang Sun, Hao Liang, Jingxuan Wei, et al.2025 · 19 citations
4. AgenticRAGTracer: A Hop-Aware Benchmark for Diagnosing Multi-Step Retrieval Reasoning in Agentic RAGFindings of the Association for Computational LinguisticsQijie You, Wenkai Yu, Hao Liang, et al.2026 · 1 citation
5. CodeFlowBench: A Multi-turn, Iterative Benchmark for Complex Code GenerationAnnual Meeting of the Association for Computational Linguistics (ACL)Sizhe Wang, Zhengren Wang, Dongsheng Ma, et al.2026 · 1 citation
6. SciFlow-Bench: Evaluating Structure-Aware Scientific Diagram Generation via Inverse ParsingAnnual Meeting of the Association for Computational Linguistics (ACL)Tong Zhang, Honglin Lin, Zhou Liu, et al.2026 · 1 citation
7. Reinforcement Learning with Semantic Rewards Enables Low-Resource Language Expansion without Alignment TaxFindings of the Association for Computational LinguisticsZeli Su, Ziyin Zhang, Zhou Liu, et al.2026
8. Rethinking Text-to-SQL: Dynamic Multi-turn SQL Interaction for Real-world Database ExplorationFindings of the Association for Computational LinguisticsLinzhuang Sun, Tianyu Guo, Hao Liang, et al.2026
9. Training Verifier to Assessing Complex Real-World Tool-Use TrajectoriesFindings of the Association for Computational LinguisticsLinzhuang Sun, Mingyang Chen, Hao Liang, et al.2026
10. The Data Frontier for Large Language Models: Selection, Synthesis, and Tools ↗Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 5: Tutorial Abstracts) · no summary yetLijun Wu, Wentao Zhang, Conghui He2026
11. Beyond Explicit Refusals: Soft-Failure Attacks on Retrieval-Augmented Generation ↗Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · no summary yetWentao Zhang et al.2026

🤝 Co-authors here

12 of 32 co-authors with a page, the most papers together first.

• Hao Liang ×4
• Bihui Yu ×3
• Jingxuan Wei ×3
• Linzhuang Sun ×3
• Bin Cui ×2
• Fan Yang ×2
• Tianpeng Li ×2
• Tianyu Guo ×2
• Zhengren Wang ×2
• Zhou Liu ×2
• Chenzheng Zhu
• Chong 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) ×8: ACL 2026 ×2, ACL 2025 ×2
• Findings of the Association for Computational Linguistics ×5: Findings of the Association for Computational Linguistics 2026 ×4

🔗 Cited by

Papers of our library cite their works 154 times (of the 149,439 papers whose bibliographies we read); 38 citing papers with a page here →.

Updated 1 November 2026

↑ back to top

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