Skip to content

Min et al. variant of in-context learning #26

Description

@dirkgr

Motivation: It's a good baseline that should be easy to implement in the catwalk context, but nobody has asked for it.

Described by Liu at al like this:
Min et al. [21] proposed ensemble ICL, where instead of using the output probability from concatenating the k training examples, the output probabilities of the model on each training example (i.e. 1-shot ICL for each of the k examples) are multiplied together. This lowers the memory cost by a factor of k/2 but increases the computational cost by a factor of 2. In terms of task performance, Min et al. [21] find that ensemble ICL outperforms the standard concatenative variant.

This depends on first getting normal few-shot ICL working on Catwalk.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions