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

Add Llama7b FSDP test for torch.compile mode #818

Merged
merged 1 commit into from
Mar 26, 2024

Conversation

pankd
Copy link
Contributor

@pankd pankd commented Mar 19, 2024

Adds a test for FSDP

@pankd pankd requested a review from regisss as a code owner March 19, 2024 13:12
@hlahkar hlahkar requested review from vivekgoe and hlahkar March 19, 2024 13:14
@regisss
Copy link
Collaborator

regisss commented Mar 19, 2024

Is it a different test from the one that was added in #761 ?

@libinta libinta added run-test Run CI for PRs from external contributors synapse 1.15 labels Mar 19, 2024
@vivekgoe
Copy link
Collaborator

@regisss This attempts to add generic support for FSDP (similar to what you already have for DeepSpeed) such that going forward any model that runs with DeepSpeed can also be run with FSDP. Also I assume that tests in this file are "slow" tests and we can add longer tests here, we wanted to add models such as Llama-70B + FSDP which take longer time here.

Test in #761 is a short test which is intended to run as part of Gaudi2 CI. Going forward if you feel there is repetition we can consider removing it.

@vivekgoe vivekgoe added run-test Run CI for PRs from external contributors and removed run-test Run CI for PRs from external contributors labels Mar 20, 2024
@vivekgoe
Copy link
Collaborator

@pankd please add a test for Llama-70B also.

@pankd pankd force-pushed the llama7b_fsdp_compile_test branch from 15f2a97 to 54c5541 Compare March 20, 2024 13:16
@vivekgoe vivekgoe added run-test Run CI for PRs from external contributors and removed run-test Run CI for PRs from external contributors labels Mar 25, 2024
@HuggingFaceDocBuilderDev

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.

@pankd pankd force-pushed the llama7b_fsdp_compile_test branch from 54c5541 to 91074d1 Compare March 26, 2024 07:09
@regisss regisss added run-test Run CI for PRs from external contributors and removed run-test Run CI for PRs from external contributors labels Mar 26, 2024
@vivekgoe
Copy link
Collaborator

@regisss we are checking last few things, will let you know when it's done.

@pankd pankd force-pushed the llama7b_fsdp_compile_test branch from 91074d1 to f828c71 Compare March 26, 2024 13:44
@vivekgoe
Copy link
Collaborator

@regisss this is ready for merge now.

@regisss regisss added run-test Run CI for PRs from external contributors and removed run-test Run CI for PRs from external contributors labels Mar 26, 2024
@regisss regisss merged commit d75064b into huggingface:main Mar 26, 2024
11 of 12 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
run-test Run CI for PRs from external contributors synapse 1.15
Projects
None yet
Development

Successfully merging this pull request may close these issues.

5 participants