A stategy utilized by a Large Language Model when generating sequences which selects the the token with the highest probability from the probability distribution without considering future tokens.
Next