Skip to content

checkpoint engine manager 'colocate' mode support npu #278

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

Draft
kevssim wants to merge 3 commits into
base: dev-swift-v5
Choose a base branch
from
Draft

checkpoint engine manager 'colocate' mode support npu #278

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
db64e71
wip
kevssim Aug 28, 2026
d6312ca
wip
kevssim Aug 28, 2026
f80caed
Merge branch 'dev-swift-v5' into dev-ckpt-engine
kevssim Aug 28, 2026
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
6 changes: 2 additions & 4 deletions src/twinkle/checkpoint_engine/manager.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -127,8 +127,6 @@ def decide_backend_engine(

platform_name = Platform.get_platform(platform).__name__
if mode == 'colocate':
if platform_name != 'GPU':
raise NotImplementedError("mode='colocate' currently requires the GPU platform.")
from twinkle.checkpoint_engine import IPCCheckpointEngine
return IPCCheckpointEngine
if mode != 'standalone':
Expand Down Expand Up @@ -166,8 +164,8 @@ def sync_weights(self, merge_and_sync=True):
self._sync_weights_naive(merge_and_sync)
return

is_master = [True] + [False] * (self.model.device_mesh.world_size - 1)
model_metadata = self.model.prepare_checkpoint_engine(is_master)
model_metadata = self.model.prepare_checkpoint_engine([True]
+ [False] * (self.model.device_mesh.world_size - 1))
self.sampler.prepare_checkpoint_engine(False)
model_kwargs, sampler_kwargs = self.backend_cls.build_topology(
self.model.device_mesh.world_size,
Expand Down
Loading