Skip to content

Minor corrections in documentation for Nucleus Sampling implementation #293

@chiamerica

Description

@chiamerica

Hello! Thank you for sharing the clean and well-documented implementation of nucleus sampling. While reading through the explanation, I noticed two small errors in the written description that could be slightly confusing for readers.
In the section that begins:
"That is, we pick the highest probable tokens until the sum of their probabilities is less that $p$."
There are two issues:

  1. Logical oversight: The condition should be "is not less than $p$" (or equivalently, "is at least $p$"), since we want the smallest set of tokens whose cumulative probability reaches or exceeds $p$.
  2. Typo: "less that" should be "less than".

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions