We propose DAA, a trainable module that enables real-time adaptation by amplifying feature-level discrepancies between known and unknown classes in TTD.

We develop DISCOVER, which enables RL agents to solve substantially more challenging tasks than previous exploration strategies in RL.