Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Fsdp grad accum monkeypatch #2064

Merged
merged 3 commits into from
Nov 16, 2024
Merged

Fsdp grad accum monkeypatch #2064

merged 3 commits into from
Nov 16, 2024

Conversation

winglian
Copy link
Collaborator

@winglian winglian commented Nov 15, 2024

Description

While we're waiting for huggingface/transformers#34645 upstream, let's patch it so users can get the other gradient accumulation fixes from 4.46.2 and have the option for FSDP + gradient accumulation.

Fixes #2039

Motivation and Context

How has this been tested?

Screenshots (if appropriate)

Types of changes

Social Handles (Optional)

@winglian
Copy link
Collaborator Author

winglian commented Nov 15, 2024

@winglian winglian merged commit d42f202 into main Nov 16, 2024
14 checks passed
@winglian winglian deleted the fsdp-grad-accum-monkeypatch branch November 16, 2024 00:11
bursteratom pushed a commit that referenced this pull request Nov 18, 2024
djsaunde pushed a commit that referenced this pull request Dec 17, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

LORA training broken on Mistral Nemo. Massive loss values immediately.
1 participant