Skip to content

Conversation

@OleBialas
Copy link
Collaborator

I have added support for array-api as described in gh-38
Now, we can use any kind of array supported by the standard.
For example, one could use CUpy and compute TRFs with GPU acceleration.
Thanks to the magic of array-api-compat this does not require any breaking API changes.

I did make some changes to private functions though and adapted the tests accordingly.
I also changed a few outdated docstrings and started adding type hints.

Copy link
Owner

@powerfulbean powerfulbean left a comment

Choose a reason for hiding this comment

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

Thank you @OleBialas for this huge updating and work!

@OleBialas OleBialas merged commit a7263d8 into powerfulbean:master Sep 9, 2025
6 checks passed
@OleBialas
Copy link
Collaborator Author

My pleasure! Maybe we could do some benchmarks to measure training with and without GPU and add a section to the docs.
By the way, I'm just returning from a retreat organized by the Max Planck Research School where several people told me they are using mTRFpy :)

@powerfulbean
Copy link
Owner

My pleasure! Maybe we could do some benchmarks to measure training with and without GPU and add a section to the docs. By the way, I'm just returning from a retreat organized by the Max Planck Research School where several people told me they are using mTRFpy :)

Thanks Ole, glad to hear this! It sounds good to me that we do some benchmarks. I was pretty busy during the past month, but I will start adding new functions following the array-API standard this week.

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

Successfully merging this pull request may close these issues.

2 participants