Skip to content
This repository was archived by the owner on Aug 1, 2024. It is now read-only.
This repository was archived by the owner on Aug 1, 2024. It is now read-only.

Struggling to Train Downstream Classifier #58

@NolanGC

Description

@NolanGC

Hi,

I'm working on training a downstream classification task from the ImageNet-22k checkpoint. When I use a TinyViT checkpoint, average over the first dimension of output and feed that into a linear classification head, the model trains appropriately. However, if I replace TinyViT with the target encoder of I-JEPA, once again averaging over the first dimension of the final layer and feeding into a linear classification head. However, the model fails to train at all in these conditions. Has anyone been able to successfully train on a downstream task?

Thank you!

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