#weights-and-biases

TPUs for Reinforcement Learning and Behavioral Modeling

How to squeeze a year of compute time into a single day.