Reliable Data for Large ML Models: Principles and Practices

This talk examines data reliability challenges in large language models and compares them with those in classic supervised machine learning systems. It uses SRE principles to recommend ways to assess and manage data risks in production.