Profiles2020-2025
Shigang Li
Beijing University of Posts and Telecommunications Beijing
7Papers
0Last author
15Unique co-authors
4First author
0Solo
First and last author come from byline position, which some venues assign alphabetically.
Recent papers7
- FlashSparse: Minimizing Computation Redundancy for Fast Sparse Matrix Multiplications on Tensor CoresAuthor 2 of 5PPoPP 2025 · 18 citations
- Near-optimal sparse allreduce for distributed deep learningAuthor 1 of 2PPoPP 2022 · 57 citations
- Efficient Quantized Sparse Matrix Operations on Tensor CoresAuthor 1 of 3SC 2022 · 27 citations
- HammingMesh: A Network Topology for Large-Scale Deep LearningAuthor 5 of 10SC 2022 · 20 citations
- Chimera: efficiently training large-scale neural networks with bidirectional pipelinesAuthor 1 of 2SC 2021 · 124 citations
- Asynchronous Decentralized SGD with Quantized and Local UpdatesAuthor 4 of 5NeurIPS 2021 · 61 citations
- Taming unbalanced training workloads in deep learning with partial collective operationsAuthor 1 of 5PPoPP 2020 · 52 citations
