Skip to content

Unable to train T5 Large #1

@espressotechie

Description

@espressotechie

I successfully trained T5 small and base models but am not able to train the large model as Colab GPU is running out of memory. I changed the batch size to 1 but still this error. Has anyone trained the T5 Large or T5 3b models? How much GPU memory will be required to train these models? Is there any workaround?

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions