Fig.1

Concept

Deferred decoding

Deferred decoding is a rollout optimization where a model predicts future latent states without reconstructing the observable output at each intermediate step, decoding only when a visible prediction is actually needed.

The rest of “Deferred decoding” is a premium feature: every concept in the library gets a precise, practitioner-focused write-up like this one, cross-linked straight from the paper summaries that use it.

Log in to unlock

← Back to the library