DocumentationArchitecture

PredictNextToken

A zoomed-in view of one operation in Piro’s stateful inference loop.

Architecture detail

PredictNextToken

Produces a causal text prediction so the observed stream can supervise fast adaptation.

← Full model
observedChunk,
state,
runtimeWeights
):
return TextHead(Forward(observedChunk, state, runtimeWeights))
← Back to architecture