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

Jump to bottom

use flash attn fuse cross entropy loss to reduce metric memory usage #2987

Draft
wants to merge 1 commit into
base: main
Choose a base branch
from
Draft

use flash attn fuse cross entropy loss to reduce metric memory usage #2987

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
6 changes: 5 additions & 1 deletion composer/metrics/nlp.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -97,7 +97,11 @@ def __init__(self, dist_sync_on_step: bool = False, ignore_index: int = -100):
super().__init__(dist_sync_on_step=dist_sync_on_step)

self.ignore_index = ignore_index
self.loss_fn = torch.nn.CrossEntropyLoss(ignore_index=ignore_index, reduction='sum')
try:
from flash_attn.losses.cross_entropy import CrossEntropyLoss as FusedCrossEntropyLoss
self.loss_fn = FusedCrossEntropyLoss(ignore_index=ignore_index, reduction='sum')
except ImportError:
self.loss_fn = torch.nn.CrossEntropyLoss(ignore_index=ignore_index, reduction='sum')
Comment on lines +100 to +104
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Should this live in llm-foundry? CC: @dakinggg

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

see above

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

i have a very old pr that i never merged

self.add_state('sum_loss', default=torch.tensor(0.), dist_reduce_fx='sum')
self.add_state('total_items', default=torch.tensor(0), dist_reduce_fx='sum')

Expand Down
Loading