The research notebook

Notes

A closer look at the questions behind the papers: how agents scale, keep learning, and make better use of experience.

Coral ink illustration of geometric forms seen through different windows

Data & generalization

Learning from a different view

Useful visual diversity depends on both the transformation and the way it enters training.

Research note · 2 min read

Teal ink illustration of branching paths that grow and reconnect

Plasticity & adaptation

Keeping visual RL able to learn

Data, the critic and the timing of updates all shape an agent's learning capacity.

Research note · 2 min read

Purple ink illustration of a sparse network opening into a larger space

Sparsity & scaling

Why sparsify RL networks?

Static sparsity offers a surprisingly simple way to improve scaling in deep RL.

Research note · 2 min read