Yuhang Li

I am a final-year PhD Candidate at Yale University, advised by Professor Priya Panda. Before my PhD study, I earned my B.S. degree from UESTC in 2020. I have completed research internships at the National University of Singapore, SenseTime, Amazon, Sony AI, and NVIDIA Research.

My research focuses on efficient inference and training of large language models, with particular expertise in low-precision approaches.

I am seeking full-time opportunities starting in 2026. If my research aligns with your team's interests, I would welcome the opportunity to connect.

Email: yuhang.li@yale.edu

News

[Sep 2025] One paper in Weight-Activation Sparsity in LLM is accepted to NeurIPS 2025.
[May 2025] Our paper, GPTAQ (arxiv), is accepted to ICML 2025. This work addressed the accumulated quantization error problem in GPTQ and only modified it with 20 lines of code.
[Apr 2024] I had a talk in CoCoSyS.
[Feb 2024] Our paper, Error-Aware ANN-SNN Conversion (journal) is accepted to IJCV. This work systematically studies the conversion error and proposes a calibration pipeline to reduce it.
[Jan 2024] I received the Baidu 2023 PhD Fellowship (10 winners/year).
[Aug 2023] I will serve as the Area Chair for EMNLP 2023.
[Oct 2022] Our Additive Power-of-Two Quantization (ICLR'20) is now supported in official PyTorch.

Projects

GPTAQ: Efficient Finetuning-Free Quantization for Asymmetric Calibration
42nd International Conference on Machine Learning (ICML 2025)
Yuhang Li, Ruokai Yin, Donghyun Lee, Ginny Xiao, Priyadarshini Panda

[PDF] [Code]

DuoGPT: Training-free Dual Sparsity through Activation-aware Pruning in LLMs
39th Conference on Neural Information Processing Systems (NeurIPS 2025).
Ruokai Yin, Yuhang Li, Donghyun Lee, Priyadarshini Panda

[PDF] [Code]

Outlier Suppression+: Accurate Quantization of Large Language Models by Equivalent and Effective Shifting and Scaling
The 2023 Conference on Empirical Methods in Natural Language Processing (EMNLP)
Xiuying Wei, Yunchen Zhang, Yuhang Li, Xiangguo Zhang, Ruihao Gong, Jinyang Guo, Xianglong Liu

[PDF] [Code]

Brecq: Pushing the Limit of Post-Training Quantization by Block Reconstruction
9th International Conference on Learning Representations (ICLR 2021)
Yuhang Li*, Ruihao Gong*, Xu Tan, Yang Yang, Peng Hu, Qi Zhang, Fengwei Yu, Wei Wang, Shi Gu

[PDF] [Code]

Additive Powers-of-Two Quantization: An Efficient Non-Uniform Discretization for Neural Networks
8th International Conference on Learning Representations (ICLR 2020)
Yuhang Li*, Xin Dong*, Wei Wang

[PDF] [Code]

Qdrop: Randomly dropping quantization for extremely low-bit post-training quantization
10th International Conference on Learning Representations (ICLR 2022)
Xiuying Wei, Ruihao Gong, Yuhang Li, Fengwei Yu, Xianglong Liu

[PDF] [Code]

Pushing the Limit of Post-Training Quantization
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, VOL. 47, NO. 7, JULY 2025
Ruihao Gong, Xianglong Liu, Yuhang Li, Yunqiang Fan, Xiuying Wei, and Jinyang Guo

[IEEE] [Code]

Error-Aware Conversion from ANN to SNN via Post-Training Parameter Calibration
International Journal of Computer Vision, 2024
Yuhang Li, Shikuang Deng, Xin Dong, Shi Gu

[Springer] [Code]

SEENN: Towards Temporal Spiking Early-Exit Neural Networks
37th Conference on Neural Information Processing Systems (NeurIPS 2023).
Yuhang Li, Tamar Geller, Youngeun Kim, Priyadarshini Panda

[PDF] [Code]

Page updated

Google Sites

Report abuse