Skip to content
This repository was archived by the owner on Mar 24, 2026. It is now read-only.
This repository was archived by the owner on Mar 24, 2026. It is now read-only.

Memory leak in training causing RAM to explode very fast #3

Description

@iSach

Hello,

I am interested in your paper which I really enjoyed reading, thanks for your great contributions and for releasing your code.

I am more familiar with PyTorch, and I have barely tried Julia in my life. I ended up running the model and outputting images, but I could never train a model for long enough before it uses extremely high amounts of RAM, causing my cluster nodes to crash.

In the image below, you can see how it quickly takes ~128 gigs of RAM in barely half an hour.

Image

Have you experienced memory issues? Is it something related with Julia preallocation for OpenBLAS or something? I am struggling to find a solution and would really like to make it work.

Thanks in advance,
Sacha Lewin.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Fields

    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