Skip to content

mark the last forward chunk in balanced vpp #74187

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 1 commit into from
Jul 24, 2025
Merged

mark the last forward chunk in balanced vpp #74187

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
13 changes: 11 additions & 2 deletions python/paddle/distributed/fleet/meta_parallel/pipeline_parallel.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1678,9 +1678,16 @@ def _store_forward_outputs(
self.output_tensors[virtual_pp_rank].pop()

def _forward_step_helper(
self, micro_dataset, micro_step, overlap_schedule_mode=False
self,
micro_dataset,
micro_step,
overlap_schedule_mode=False,
check_is_last_chunk=False,
):
virtual_pp_rank = self._get_virtual_pp_rank(micro_step, forward=True)
if check_is_last_chunk and virtual_pp_rank == self.num_model_chunks - 1:
os.environ["FLAGS_last_vpp_chunk_forward"] = "1"

self.set_virtual_pipeline_rank(virtual_pp_rank)

input_tensor = self._get_forward_input(virtual_pp_rank)
Expand Down Expand Up @@ -3246,7 +3253,9 @@ def forward_backward_pipeline(

self._record_stamp("F", forward_micro_step_id, '"B"', forward=True)
output_tensor = self._forward_step_helper(
micro_dataset, forward_micro_step_id
micro_dataset,
forward_micro_step_id,
check_is_last_chunk=True,
)
self._record_stamp("F", forward_micro_step_id, '"E"', forward=True)

Expand Down
Loading