Skip to content
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

Avoid init_ddp for inference #12011

Open
wants to merge 6 commits into
base: main
Choose a base branch
from
+26 −14
Open

Avoid init_ddp for inference #12011

Show file tree
Hide file tree
Changes from 2 commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
e527dfd
Avoid init_ddp for inference
athitten Jan 31, 2025
d637242
Apply isort and black reformatting
athitten Jan 31, 2025
249d49f
Remove unused import
athitten Feb 4, 2025
94f17a0
More Pylint fixes
athitten Feb 5, 2025
1d29e18
Fix string formatting
athitten Feb 5, 2025
a2a9ba9
Another line too long fix
athitten Feb 5, 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
5 changes: 4 additions & 1 deletion nemo/lightning/megatron_parallel.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -42,6 +42,7 @@

import torch
import torch.distributed
from lightning.pytorch.trainer.states import TrainerFn
from lightning.pytorch.utilities import move_data_to_device
from megatron.core import parallel_state
from megatron.core.distributed import DistributedDataParallel as McoreDDP
Expand Down Expand Up @@ -564,7 +565,9 @@ def init_model_parallel(self):
if self.convert_module_fn:
self.apply_convert_module_fn()

self.init_ddp()
# Skip init_ddp for inference i.e testing as it can lead to OOM.
if not self.trainer.state.fn == TrainerFn.TESTING:
self.init_ddp()

def apply_convert_module_fn(self):
for i in range(len(self)):
Expand Down
6 changes: 4 additions & 2 deletions nemo/lightning/pytorch/strategies/megatron_strategy.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -477,11 +477,13 @@ def setup_megatron_parallel(self, trainer: pl.Trainer) -> None:
convert_module_fn=convert_module_fn,
)

# Assign trainer to megatron_parallel before init_model_parallel as its required to check stage of trainer
# (TESTING or not) in init_model_parallel.
self.megatron_parallel.trainer = trainer

if self._init_model_parallel:
self.init_model_parallel()

self.megatron_parallel.trainer = trainer

# check signature-def of self.model.configure_optimizers to check if there's an optional arg: megatron_parallel
sig = inspect.signature(self.model.configure_optimizers)
if "megatron_parallel" in sig.parameters:
Expand Down
Loading