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

Feature/kan experiment #999

Open
wants to merge 16 commits into
base: main
Choose a base branch
from
Open

Feature/kan experiment #999

wants to merge 16 commits into from

Conversation

marcopeix
Copy link
Contributor

Add KAN model
Benchmark of KAN on M3 and M4 dataset against MLP and NBEATS

Copy link

Check out this pull request on  ReviewNB

See visual diffs & provide feedback on Jupyter Notebooks.


Powered by ReviewNB

@marcopeix marcopeix marked this pull request as ready for review May 10, 2024 13:58
@marcopeix marcopeix requested a review from cchallu May 10, 2024 13:58
@AzulGarza AzulGarza self-requested a review May 10, 2024 19:48
Copy link
Member

@AzulGarza AzulGarza left a comment

Choose a reason for hiding this comment

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

very cool @marcopeix, i left a couple of comments. also i think it might be pretty awesome to release the Auto version as well. perhaps in a different pr, once this one is merged.

experiments/kan_benchmark/README.md Outdated Show resolved Hide resolved
neuralforecast/models/kan.py Show resolved Hide resolved
experiments/kan_benchmark/environment.yml Outdated Show resolved Hide resolved
@AzulGarza AzulGarza requested a review from elephaint May 10, 2024 19:52
@elephaint elephaint linked an issue May 13, 2024 that may be closed by this pull request
Copy link
Contributor

@elephaint elephaint left a comment

Choose a reason for hiding this comment

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

Cool!

  • I think we should also add an AutoKANLinear model in models.ipynb, I think users will like the ability to have it auto-optimize when they are playing with the architecture. You can just copy-paste most of the code from a comparable MLP-based architecture and change the names.
  • I think KANLinear should also be added to models.py and evaluation.py in action_files/test_models/src

@marcopeix marcopeix requested a review from cchallu May 15, 2024 18:51
@cchallu cchallu self-requested a review May 15, 2024 23:06
@cchallu
Copy link
Contributor

cchallu commented May 15, 2024

@marcopeix we also need to add KAN to the evaluation pipeline in https://github.com/Nixtla/neuralforecast/tree/main/action_files. You can check with Olivier how to add it,

Copy link
Contributor

@elephaint elephaint left a comment

Choose a reason for hiding this comment

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

Looks good, @marcopeix!

@marcopeix
Copy link
Contributor Author

@cchallu I already added KAN to the evaluation pipeline in action_files. Olivier checked it and says it's good! Let me know if I am missing something!

@cchallu cchallu requested a review from AzulGarza May 21, 2024 17:36
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Projects
None yet
Development

Successfully merging this pull request may close these issues.

when to surport KANS?
4 participants