Skip to content

small update on rotary embedding #9354

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 6 commits into from
Sep 3, 2024
Merged

small update on rotary embedding #9354

Changes from all commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
8670ec7
update
yiyixuxu Sep 3, 2024
6895f9c
Merge branch 'main' into small-update-rotary
yiyixuxu Sep 3, 2024
29275cd
fix
yiyixuxu Sep 3, 2024
41d3210
Merge branch 'small-update-rotary' of github.com:huggingface/diffuser…
yiyixuxu Sep 3, 2024
769c538
Merge branch 'main' into small-update-rotary
yiyixuxu Sep 3, 2024
88e1c94
Merge branch 'main' into small-update-rotary
sayakpaul Sep 3, 2024
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
7 changes: 5 additions & 2 deletions src/diffusers/models/embeddings.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -608,8 +608,11 @@ def get_1d_rotary_pos_embed(
pos = torch.from_numpy(pos) # type: ignore # [S]

theta = theta * ntk_factor
freqs = 1.0 / (theta ** (torch.arange(0, dim, 2, dtype=freqs_dtype)[: (dim // 2)] / dim)) / linear_factor # [D/2]
freqs = freqs.to(pos.device)
freqs = (
Copy link
Collaborator Author

Choose a reason for hiding this comment

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

@sayakpaul
this is the only code change from #9321
Copy link
Member

Choose a reason for hiding this comment

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

(nit): We could add a comment on emphasizing the impact of creating torch.arange() on device so as to prevent syncs. This will be helpful references for us going forward. What do you think?

We could do similar for the changes introduced in https://github.com/huggingface/diffusers/pull/9307/files as well.
Copy link
Collaborator Author

Choose a reason for hiding this comment

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

this did not cause the sync, though
1.0
/ (theta ** (torch.arange(0, dim, 2, dtype=freqs_dtype, device=pos.device)[: (dim // 2)] / dim))
/ linear_factor
) # [D/2]
freqs = torch.outer(pos, freqs) # type: ignore # [S, D/2]
if use_real and repeat_interleave_real:
# flux, hunyuan-dit, cogvideox
Expand Down
Loading