About Me

I am now a first-year PhD candidate at National University of Singapore (NUS), supervised by Dr. Michael Qizhe Shieh. My research is driven by the passion for advancing the edge intelligence of large language models (LLMs). I am particularly focused on the paradigm shift from deep to long learning, aiming to enhance LLM capabilities by learning effectively from substantial context. My recent work concentrates on improving LLMs’ in-depth understanding of complex context, specifically targeting:

• Effective and efficient long-context modeling;
• Unified understanding of multimodal context.

News

• May 2025: RAPID has been accepted to ICML 2025 as Spotlight!
• Feb 2025: We release the LongPO, a self-evolving long-context LLM training approach for both context extension and long-context alignment in one stage without external annotation.
• Feb 2025: LongPO has been accepted to ICLR 2025!
• Jan 2024: CLEX has been accepted to ICLR 2024!
• Oct 2023: We release the CLEX, a length extrapolation method that enables LLMs to access the context length up to 4x~8x the training length!

Publications

• RAPID: Long-Context Inference with Retrieval-Augmented Speculative Decoding
  Guanzheng Chen*, Qilong Feng*, Jinjie Ni, Xin Li, Michael Qizhe Shieh.
  The Forty-Second International Conference on Machine Learning (ICML’25, Spotlight)

  PDF Code

• LongPO: Long Context Self-Evolution of Large Language Models through Short-to-Long Preference Optimization
  Guanzheng Chen, Xin Li, Michael Qizhe Shieh, Lidong Bing.
  The Thirteenth International Conference on Learning Representations (ICLR’25)

  PDF Code

• VCD: Mitigating Object Hallucinations in Large Vision-Language Models through Visual Contrastive Decoding
  Sicong Leng, Hang Zhang, Guanzheng Chen, Xin Li, Shijian Lu, Chunyan Miao, Lidong Bing.
  The IEEE/CVF Conference on Computer Vision and Pattern Recognition 2024 (CVPR’24)

  PDF Code

• CLEX: Continuous Length Extrapolation for Large Language Models
  Guanzheng Chen, Xin Li, Zaiqiao Meng, Shangsong Liang, Lidong Bing.
  The Twelfth International Conference on Learning Representations (ICLR’24)

  PDF Code

• Revisiting Parameter-Efficient Tuning: Are We Really There Yet?
  Guanzheng Chen, Fangyu Liu, Zaiqiao Meng, Shangsong Liang.
  The 2022 Conference on Empirical Methods in Natural Language Processing (EMNLP’22, Oral Presentation).

  PDF Code

• Multi-Relational Graph Representation Learning with Bayesian Gaussian Process Network
  Guanzheng Chen, Jinyuan Fang, Zaiqiao Meng, Qiang Zhang, Shangsong Liang.
  Thirty-Sixth AAAI Conferene on Artificial Intelligence (AAAI’22).
  

  PDF Code

Services

• Conference reviewer: Neurips 2024, ICLR 2025, ICML 2025, ACL Roling Review

• Journal reviewer: Neurocomputing, IEEE Transactions on Pattern Analysis and Machine Intelligence