Thanks to visit codestin.com
Credit goes to github.com

Skip to content

[Bug]: Flux2 full finetuning only works without offloading #1274

Description

@dxqb

What happened?

Flux2 full finetuning fails if offloading is enabled, see error log below.

As a workaround if you need full finetuning, install #1279

What did you expect would happen?

Relevant log output

RuntimeError: Trying to backward through the graph a second time (or directly access saved tensors after they have already been freed). Saved intermediate values of the graph are freed when you call .backward() or autograd.grad(). Specify retain_graph=True if you need to backward through the graph a second time or if you need to access saved tensors after calling backward.

Generate and upload debug_report.log

No response

Metadata

Metadata

Assignees

No one assigned

    Labels

    bugSomething isn't workingwaitingwaiting for something / cannot fix currently

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions