Top-p Sampling
Sampling only from the smallest set of tokens whose probabilities reach p.
Aperçu
Top-p, or nucleus sampling, keeps the most likely tokens until their cumulative probability reaches a threshold, then samples from that set. It solves the problem that a fixed top-k includes nonsense when the distribution is peaked and excludes reasonable options when it is flat. It exists because the set of plausible continuations should be sized by the model's own confidence.
Graphe des technologies
Comment Top-p Sampling se rattache au reste de la pile.
Alternatives
-
Beam Search
technique
Une alternative à Top-p Sampling · source
Comparaisons
Comparaisons détaillées avec les alternatives ci-dessus.