Skip to content

other

Hidden Technical Debt in Machine Learning Systems

A Google research paper arguing that data dependencies, testing, monitoring and supporting infrastructure, not model code, dominate the long-term cost of production machine learning.

Current clusters