Skip to content

[Llama4] Enable tp comm overlap for llama4 #13940

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.

Sign up for GitHub

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

Merged
merged 2 commits into from
Jun 17, 2025
Merged

[Llama4] Enable tp comm overlap for llama4 #13940

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
c7635fd
pass in the arg
Jun 16, 2025
7659b86
Merge branch 'main' into gaod/llama4/tp_comm_overlap_vlm
gdengk Jun 16, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions nemo/collections/vlm/llama4/model/base.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -161,6 +161,7 @@ def configure_model(self, tokenizer, vp_stage: Optional[int] = None) -> "MCoreNe
# set token_drop setting from config
self.language_transformer_config.moe_pad_expert_input_to_capacity = self.moe_pad_expert_input_to_capacity
self.language_transformer_config.moe_expert_capacity_factor = self.moe_expert_capacity_factor
self.language_transformer_config.tp_comm_overlap = self.tp_comm_overlap

# During fake lightning initialization, pass 0 to bypass the assertion that vp_stage must be
# non-None when using virtual pipeline model parallelism
Expand Down
Loading