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

Update DS config to align with recommended settings #1730

Merged
merged 1 commit into from
Feb 5, 2025

Conversation

ckvermaAI
Copy link
Contributor

Update Deepspeed-zero1 config to align with recommended settings (https://docs.habana.ai/en/latest/PyTorch/DeepSpeed/DeepSpeed_User_Guide/DeepSpeed_User_Guide.html#integrating-deepspeed-with-gaudi)

Configure the throughput timer to be unsynchronized
Configure contiguous_gradients=false

@yafshar
Copy link
Contributor

yafshar commented Jan 28, 2025

@ckvermaAI thanks. Do we need to update the rest of the configs? For example,

has "contiguous_gradients": false, while the reference has different recommendation.

@ckvermaAI
Copy link
Contributor Author

@yafshar Right now, we do not need to update the rest of the DS configs.

@libinta libinta added the run-test Run CI for PRs from external contributors label Feb 3, 2025
@regisss regisss changed the base branch from main to synapse_1_20 February 5, 2025 10:50
Copy link
Collaborator

@regisss regisss left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM!

@regisss regisss merged commit d214819 into huggingface:synapse_1_20 Feb 5, 2025
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 synapse1.20
Projects
None yet
Development

Successfully merging this pull request may close these issues.

5 participants