Chenggang Zhao's picture

Chenggang Zhao

LyricZ

·

https://github.com/LyricZhao

LyricZhao

AI & ML interests

Building efficient machine learning systems.

Recent Activity

new activity 5 days ago

deepseek-ai/DeepSeek-V4-Pro:关于 "Observations and Proposals" 中激活函数建议的疑问：去掉 gate projection 为何能放宽 EP 带宽要求？

liked a model 5 days ago

deepseek-ai/DeepSeek-V4-Pro

authored a paper 4 months ago

mHC: Manifold-Constrained Hyper-Connections

View all activity

Organizations

authored a paper 4 months ago

mHC: Manifold-Constrained Hyper-Connections

Paper • 2512.24880 • Published Dec 31, 2025 • 323

authored a paper 12 months ago

Insights into DeepSeek-V3: Scaling Challenges and Reflections on Hardware for AI Architectures

Paper • 2505.09343 • Published May 14, 2025 • 76

authored 2 papers over 1 year ago

DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning

Paper • 2501.12948 • Published Jan 22, 2025 • 448

Auxiliary-Loss-Free Load Balancing Strategy for Mixture-of-Experts

Paper • 2408.15664 • Published Aug 28, 2024 • 15