🧑‍🔬 Lefei Zhang

🏛️ Affiliations

🎓Wuhan UniversityWuhan · China2009–2026 · 250 papersEducation · PhDEmployment since 2017 · Professor, School of Computer Sciencecurrent🔬State Key Laboratory of Information Engineering in Surveying Mapping and Remote SensingWuhan · China2009–2026 · 72 paperscurrent🔬Henan Academy of SciencesZhengzhou · China2024–2026 · 7 paperscurrent🎓Hong Kong Polytechnic UniversityHong Kong · China2016–2022 · 13 papers🔬State Key Laboratory of Software EngineeringHubei2015–2018 · 10 papers🎓Aerospace Information Research InstituteBeijing · China2024–2025 · 4 papers

From the bylines of their papers (OpenAlex, 2026-10-14), ORCID and Wikidata; institutions on papers are matched by machine. 7 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. Intention Analysis Makes LLMs A Good Jailbreak Defender (2025 · 21 citations)
   Safety-aligned LLMs can still be broken by jailbreaks, which hide a harmful request inside elaborate role-play or fictional framings.
2. SpindleKV: A Novel KV Cache Reduction Method Balancing Both Shallow and Deep Layers (2025 · 9 citations)
   Long-context inference with LLMs is limited by the GPU memory consumed by the KV cache. Earlier token eviction, merging and quantization methods compress deep layers well, but shallow layers resist them. In a preliminary analysis of LLaMA2-7B-chat on 2WikiMQA, the authors found two patterns. Deep layers concentrate…
3. KV-Latent: Dimensional-level KV Cache Reduction with Frequency-aware Rotary Positional Embedding (2025 · 6 citations)
   Decoder-only LLMs store keys and values for every generated token. This KV Cache grows with sequence length and becomes a major limit on memory and bandwidth during inference. KV-Latent tackles this at the level of individual head dimensions, a direction earlier head-, layer- and token-level methods have left largely…
4. RACER: Retrieval-Augmented Contextual Rapid Speculative Decoding (2026 · 1 citation)
   Large language models decode one token at a time, which makes inference slow. Speculative decoding speeds this up by drafting several tokens and checking them in parallel with the target model. Training-free variants have two weaknesses. Retrieval-based drafters fail when the context contains no exact match.…
5. From AR to Diffusion: Efficiently Adapting Large Language Models with Strictly Causal and Elastic Horizons (2026)
   Diffusion language models can produce several tokens at once. However, they usually depend on bidirectional attention, which conflicts with how pre-trained autoregressive (AR) models such as GPT-style LLMs are built. Because of this, they are often trained from scratch at great cost. Existing block diffusion methods…

All works

7 works of theirs in our library, the most cited first; 5 with a page here, the others link their original until we summarise them.

1. Progressive Reconstruction of Visual Structure for Image Inpainting ↗IEEE/CVF International Conference on Computer Vision (ICCV) · no summary yetJingyuan Li et al.2019 · 183 citations
2. Self-Ensembling Attention Networks: Addressing Domain Shift for Semantic Segmentation ↗Proceedings of the AAAI Conference on Artificial Intelligence · no summary yetYonghao Xu et al.2019 · 69 citations
3. Intention Analysis Makes LLMs A Good Jailbreak DefenderInternational Conference on Computational Linguistics (COLING)Yuqi Zhang, Liang Ding, Lefei Zhang, et al.2025 · 21 citations
4. SpindleKV: A Novel KV Cache Reduction Method Balancing Both Shallow and Deep LayersAnnual Meeting of the Association for Computational Linguistics (ACL)Zicong Tang, Shi Luohe, Zuchao Li, et al.2025 · 9 citations
5. KV-Latent: Dimensional-level KV Cache Reduction with Frequency-aware Rotary Positional EmbeddingAnnual Meeting of the Association for Computational Linguistics (ACL)Shi Luohe, Zuchao Li, Lefei Zhang, et al.2025 · 6 citations
6. RACER: Retrieval-Augmented Contextual Rapid Speculative DecodingFindings of the Association for Computational LinguisticsZihong Zhang, Zuchao Li, Lefei Zhang, et al.2026 · 1 citation
7. From AR to Diffusion: Efficiently Adapting Large Language Models with Strictly Causal and Elastic HorizonsAnnual Meeting of the Association for Computational Linguistics (ACL)Xiangyu Ma, Teng Xiao, Zuchao Li, et al.2026

🤝 Co-authors here

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

• Zuchao Li ×4
• Baoyuan Qi ×2
• Hai Zhao ×2
• Liu Guoming ×2
• Ping Wang ×2
• Shi Luohe ×2
• Dacheng Tao
• Liang Ding
• Teng Xiao

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, ACL 2025 ×2
• International Conference on Computational Linguistics (COLING): COLING 2025
• Findings of the Association for Computational Linguistics: Findings of the Association for Computational Linguistics 2026
• IEEE/CVF International Conference on Computer Vision (ICCV)
• Proceedings of the AAAI Conference on Artificial Intelligence

🔗 Cited by

Papers of our library cite their works 34 times (of the 149,439 papers whose bibliographies we read); 9 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/lefei-zhang/