Skip to content

Conversation

@daniellepintz
Copy link
Contributor

@daniellepintz daniellepintz commented Oct 30, 2025

Test Plan:

python -m apps.sft.main --config apps/sft/deepseek_v3.yaml

@meta-cla meta-cla bot added the CLA Signed This label is managed by the Meta Open Source bot. label Oct 30, 2025
Comment on lines +38 to +39
data_parallel_shard_degree: -1
tensor_parallel_degree: 1
Copy link
Contributor

Choose a reason for hiding this comment

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

Based on a conversation with @tianyu-l this should work for 32 nodes

Suggested change
data_parallel_shard_degree: -1
tensor_parallel_degree: 1
data_parallel_shard_degree: -1
tensor_parallel_degree: 8

It also may be worth looking at some of the runs here for reference.

Not sure of the latest status of ETP in titan but it'd be worth confirming that as well.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

CLA Signed This label is managed by the Meta Open Source bot.

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants