Profiles2024-2026
Xinglin Pan
The Hong Kong University of Science and Technology (Guangzhou)
9Papers
0Last author
20Unique co-authors
2First author
0Solo
First and last author come from byline position, which some venues assign alphabetically.
Recent papers9
- HierMoE: Accelerating MoE Training with Hierarchical Token Deduplication and Expert SwapAuthor 2 of 6INFOCOM 2026 · 7 citations
- ZipCCL: Efficient Lossless Data Compression of Communication Collectives for Accelerating LLM TrainingAuthor 2 of 5SIGCOMM 2026 · 1 citation
- Compass: Dissecting Communication and Computation Operators for Efficient LLM TrainingAuthor 4 of 6INFOCOM 2026
- ZipServ: Fast and Memory-Efficient LLM Inference with Hardware-Aware Lossless CompressionAuthor 3 of 8ASPLOS 2026
- ParZC: Parametric Zero-Cost Proxies for Efficient NASAuthor 3 of 8AAAI 2025 · 14 citations
- FSMoE: A Flexible and Scalable Training System for Sparse Mixture-of-Experts ModelsAuthor 1 of 8ASPLOS 2025 · 12 citations
- Pruner-Zero: Evolving Symbolic Pruning Metric From Scratch for Large Language ModelsAuthor 5 of 7ICML 2024 · 64 citations
- Should We Really Edit Language Models? On the Evaluation of Edited Language ModelsAuthor 5 of 6NeurIPS 2024 · 25 citations
- Parm: Efficient Training of Large Sparsely-Activated Models with Dedicated SchedulesAuthor 1 of 6INFOCOM 2024 · 13 citations
