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.

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

Onnx runtime in arm64 docker image #23673

Closed
danitico opened this issue Aug 16, 2022 · 3 comments
Closed

Onnx runtime in arm64 docker image #23673

danitico opened this issue Aug 16, 2022 · 3 comments
Assignees

Comments

@danitico
Copy link

Describe the bug
When working with the arm64 image, I am having problems when trying to use onxxruntime to infer embeddings. In the screenshot section I show the error. This bug is originated in the onxxruntime library but it has been recently fixed in microsoft/onnxruntime#12335

To Reproduce
Steps to reproduce the behavior:

  1. Use BertBaseEmbedder in an arm64 environment

Expected behavior
It should load the onnxruntime correctly.

Screenshots

Screenshot 2022-08-16 at 09 52 29

Environment:

  • OS: macOS M1
  • Infrastructure: self-hosted
  • Version: 12.5

Vespa version
I am using the version 8.31.22

Additional context
Add any other context about the problem here.

@bratseth bratseth self-assigned this Aug 16, 2022
@bratseth
Copy link
Member

Thanks! #23674

This will be available in 8.36, available Thursday if all goes normally.

@danitico
Copy link
Author

@bratseth I have tried the new version of the docker image and I encountered the same error. I have been checking the directories and in /opt/vespa-deps/lib64 are located the onnxruntime libraries. I have copied those ones to /opt/vespa/lib64 and everything is working again!

I think the directory /opt/vespa-deps/lib64 should also be added to java.library.path

@bratseth
Copy link
Member

This didn't fully make it into 8.36 after all, you'll need 8.38. Sorry for the delay.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants