- 👋 Hi, I’m @Zishan-Shao (You can call me Bruce, I know it's hard to pronounce this)
- 👀 I’m broadly interested in Large Language Models (LLMs) and ML Systems: spanning LLM efficiency (model compression, hardware-aware acceleration) and mechanistic analysis (decode-time dynamics).
- 🌱 I’m currently working on decode-time evaluation methodologies and efficient inference serving.
- 💞️ I’m looking to collaborate on hardware-efficient LLM serving or mechanistic interpretability.
- 🏋️ Hobbies: Competitive Powerlifting (100kg+ BP / 160kg+ SQ & DL), Combat Sports (MMA, Boxing, BJJ 🥋), and occasionally exploring Teyvat in Genshin Impact.
- 📫 How to reach me: bruceshao5418@gmail.com, 15801067888@163.com
-
@duke University
- Durham
-
17:13
(UTC -04:00) - https://zishan-shao.github.io/
- https://orcid.org/0009-0003-7873-8857
Pinned Loading
-
decodeshare
decodeshare Public🏆[ICML 2026 Spotlight] Official implementation of "DecodeShare: Tracing the Shared Subspace of LLM Decode-Time Decisions"
Python 4
-
lowrankarena
lowrankarena PublicFirst standardized benchmark for SVD-compressed LLMs up to 70B, with 1,000+ GPU hours and 5TiB+ reproduced open checkpoints.
Python 2
-
zeus-mm
zeus-mm PublicForked from Ting-Justin-Jiang/ZEUS
[ACM MM 2026]⚡ZEUS accelerates your diffuser. Any modality. Any model. Any scheduler. https://yixiao-wang-stats.github.io/zeus/
Python
-
sada-icml
sada-icml PublicForked from Ting-Justin-Jiang/sada-icml
[ICML 2025] Official Repo for Stability-guided Adaptive Diffusion Acceleration. 🚀🌙Accelerating off-the-shelf diffusion model with a unified stability criterion.
Python
-
If the problem persists, check the GitHub status page or contact support.
