Get the latest tech news

Reproducing DeepSeek's MHC: When Residual Connections Explode


Taylor Kolasinski - Engineering at FlowMode. ML systems & research, reinforcement learning, robotics. Based in Brooklyn, NY.

None

Get the Android app

Or read this on Hacker News

Read more on:

Photo of MHC

MHC

Photo of reproducing deepseek

reproducing deepseek

Photo of residual connections

residual connections

Related news:

News photo

MHC: Manifold-Constrained Hyper-Connections