Skip to content

Keep FSDP's and DDP's finish_grad_sync API identical #3070

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
chtruong814 merged 1 commit into from
Jan 24, 2026
Merged

Keep FSDP's and DDP's finish_grad_sync API identical #3070

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
5 changes: 4 additions & 1 deletion megatron/core/distributed/fsdp/src/megatron_fsdp/megatron_fsdp.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1105,7 +1105,7 @@ def attach_grad_to_optimizer_state(self):
"""
self.param_and_grad_buffer.update_main_grads()

def finish_grad_sync(self):
def finish_grad_sync(self, force_all_reduce: Optional[bool] = False):
"""
Finishes grad sync (all-reduce or reduce-scatter) communication operations
for all model gradients. Call prior to the optimization step to resolve
Expand All @@ -1114,6 +1114,9 @@ def finish_grad_sync(self):
When overlap_grad_reduce is set to True, waits for asynchronous communication
calls to complete. When overlap_grad_reduce is set to False, calls synchronous
communication ops.

NOTE: force_all_reduce is included as an argument to maintain API compatibility
with DDP.force_grad_sync.
"""
# Synchronize gradient reduce-scatter operations for all model gradients.
self.synchronize_gradient_reduce()
Expand Down
Loading