Scholar
Pingzhi Li
Google Scholar ID: QUfhEyQAAAAJ
Ph.D. student @UNC-Chapel Hill
Citations & Impact
Citations
188
H-index
5
i10-index
3
Publications
20
Co-authors
20
list available
Publications
Resume (English only)
Academic Achievements
- 1. Mozart: Modularized and Efficient MoE Training on 3.5D Wafer-Scale Chiplet Architectures (NeurIPS 2025, Spotlight)
- 2. Occult: Optimizing Collaborative Communication across Experts for Accelerated Parallel MoE Training and Inference (ICML 2025)
- 3. Advancing MoE Efficiency: A Collaboration-Constrained Routing (C2R) Strategy for Better Expert Parallelism Design (NAACL 2025, SAC Award)
- 4. Model-GLUE: Democratized LLM Scaling for A Large Model Zoo in the Wild (NeurIPS 2024)
- 5. Revisiting Zeroth-Order Optimization for Memory-Efficient LLM Fine-Tuning: A Benchmark (ICML 2024)
- 6. Merge, Then Compress: Demystify Efficient SMoE with Hints from Its Routing Policy (ICLR 2024, Spotlight)
- Awards:
- - 2025 NAACL Low Resource Methods for NLP SAC Award
- - 1st Place of ACM/IEEE Quantum Computing for Drug Discovery Challenge 2023
- - Outstanding Graduates Scholarship, USTC 2023
- - Silver Medal in Kaggle Feedback Prize - Evaluating Student Writing 2022
- - Outstanding Student Scholarship, USTC 2020/21/22
Research Experience
- Currently a part-time research intern at Apple AI/ML, working with the Foundation Model team. Previously worked closely with Dr. Hanrui Wang.
Education
- Ph.D. in Computer Science at UNC-Chapel Hill, advised by Prof. Tianlong Chen; B.E. in CS from USTC.
Background
- Research interests: efficient AI computing, ML for science and vice versa. Enjoys simple and useful things, building systems, and turning thoughts into reality.
Miscellany
- Personal interests include browsing Twitter, Last.fm, Instagram, and writing blogs.
