Skip to content

Update cpu check to have better warning #3757

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

Closed
wants to merge 5 commits into from
Closed

Update cpu check to have better warning #3757

Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
0be93c1
update cpu check
Jan 30, 2025
f53e084
Merge branch 'main' into better-mixed-init-warning
Jan 31, 2025
893fb2c
test
Feb 4, 2025
a798e1c
afterload
Feb 6, 2025
fb4bb89
precommit
Feb 6, 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
9 changes: 4 additions & 5 deletions composer/callbacks/memory_monitor.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -129,12 +129,11 @@ def __init__(
self.memory_keys = memory_keys
self.dist_aggregate_batch_interval = dist_aggregate_batch_interval

def init(self, state: State, logger: Logger) -> None:
def after_load(self, state: State, logger: Logger) -> None:
# Not relying on `torch.cuda.is_available()` since the model could be on CPU.
model_device = next(state.model.parameters()).device

if model_device.type not in ('cuda', 'meta'):
warnings.warn(f'The memory monitor only works on CUDA devices, but the model is on {model_device.type}.')
devices = {p.device.type for p in state.model.parameters()}
if 'cuda' not in devices and 'meta' not in devices:
warnings.warn(f'The memory monitor only works on CUDA devices, but the model is on {devices}.')

def after_train_batch(self, state: State, logger: Logger):
memory_report = {}
Expand Down
Loading