-
Notifications
You must be signed in to change notification settings - Fork 27.3k
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.
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
Fix position embeddings singular/plural #33678
Conversation
cc @amyeroberts for review! |
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
@amyeroberts sorry for the ping, just keeping this one higher in the notifications - hub tests were just flaky :) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks for fixing!
Found out about an issue in |
* fix position embeddings * [run-slow] blip, blip_2, instructblip, instructblipvideo * fix init * [run-slow] blip, blip_2, instructblip, instructblipvideo * fix copies * [run-slow] blip, blip_2, instructblip, instructblipvideo * [run-slow] blip, blip_2, instructblip, instructblipvideo * handle exception where list + tensors are cat'd * [run-slow] blip, blip_2, instructblip, instructblipvideo * add missing default * [run-slow] blip, blip_2, instructblip, instructblipvideo
* fix position embeddings * [run-slow] blip, blip_2, instructblip, instructblipvideo * fix init * [run-slow] blip, blip_2, instructblip, instructblipvideo * fix copies * [run-slow] blip, blip_2, instructblip, instructblipvideo * [run-slow] blip, blip_2, instructblip, instructblipvideo * handle exception where list + tensors are cat'd * [run-slow] blip, blip_2, instructblip, instructblipvideo * add missing default * [run-slow] blip, blip_2, instructblip, instructblipvideo
* fix position embeddings * [run-slow] blip, blip_2, instructblip, instructblipvideo * fix init * [run-slow] blip, blip_2, instructblip, instructblipvideo * fix copies * [run-slow] blip, blip_2, instructblip, instructblipvideo * [run-slow] blip, blip_2, instructblip, instructblipvideo * handle exception where list + tensors are cat'd * [run-slow] blip, blip_2, instructblip, instructblipvideo * add missing default * [run-slow] blip, blip_2, instructblip, instructblipvideo
* fix position embeddings * [run-slow] blip, blip_2, instructblip, instructblipvideo * fix init * [run-slow] blip, blip_2, instructblip, instructblipvideo * fix copies * [run-slow] blip, blip_2, instructblip, instructblipvideo * [run-slow] blip, blip_2, instructblip, instructblipvideo * handle exception where list + tensors are cat'd * [run-slow] blip, blip_2, instructblip, instructblipvideo * add missing default * [run-slow] blip, blip_2, instructblip, instructblipvideo
What does this PR do?
It seems a few Blip-derived models have both
self.position_embedding
andself.position_embeddings
, plural, defined in the same module and causing some errors. This aims at fixing those. Noticed it in #31368 .Causes many things to break because self.position_embedding isn't defined since it's self.position_embeddings (with an s) instead.