Training Procedure Insights
Rosanne and Lukas discuss modifying the training procedure to restrict movement in a fixed loss landscape. They explore how changing a smaller set of numbers affects the network through linear transformation, shedding light on the impact of network size on tasks like mnist and Cifor.In this clip
From this podcast
Gradient Dissent - A Machine Learning Podcast
Rosanne Liu — Conducting Fundamental ML Research as a Nonprofit
Related Questions