LlamaOmniAlignment

Enums · AiDotNet.TextToSpeech.MultiModal

Which hidden state stands for a response token when the speech decoder trains.

Fields

Field Summary
LlamaOmniAlignment Paper The paper’s c_i = M(P(S), Y_<i): the state that predicts y_i, the one inference reads.
LlamaOmniAlignment Release The release’s training code: the state at y_i’s own position (the one that has read it), one position later than inference reads.
int value__