Skip to content

Match setfit.Trainer Ray compatibility with Transformers #588

Open
@zbloss

Description

@zbloss

Currently the setfit.Trainer class only accepts datasets.Dataset-like objects, which means I cannot pass ray.data Dataset like I would normally do when using transformers.Trainer

See ray docs for example code: https://docs.ray.io/en/latest/train/getting-started-transformers.html

Currently, passing a ray dataset fails at this line:

column_names = set(dataset.column_names)

Is it possible to allow SetFit to run on ray using a distributed dataset?

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