Skip to content

update get_parameter_local_cp#22

Merged
Jintao-Huang merged 1 commit intomodelscope:mainfrom
Jintao-Huang:update_get_parameter_local_cp
Apr 9, 2026
Merged

update get_parameter_local_cp#22
Jintao-Huang merged 1 commit intomodelscope:mainfrom
Jintao-Huang:update_get_parameter_local_cp

Conversation

@Jintao-Huang
Copy link
Copy Markdown
Collaborator

No description provided.

Copy link
Copy Markdown

@gemini-code-assist gemini-code-assist bot left a comment

Choose a reason for hiding this comment

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

Code Review

This pull request introduces a local implementation of the get_parameter_local_cp function in gated_delta_net.py, sourced from Megatron-LM, and updates the GatedDeltaNet forward pass to use this local version. Feedback suggests adding an assertion to ensure the parameter dimension is divisible by the context parallel size, which prevents potential failures in collective operations due to non-uniform tensor shapes.

@Jintao-Huang Jintao-Huang merged commit 4636e98 into modelscope:main Apr 9, 2026
1 check passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant