# Rainbow

Combines prioritized replay, multi-step returns, distributional values, noisy networks, double and dueling learning in a single agent.

## What it is used for

A strong value-based baseline for discrete-action reinforcement learning.