search

NIEHS / beethoven

BEETHOVEN is: Building an Extensible, rEproducible, Test-driven, Harmonized, Open-source, Versioned, ENsemble model for air quality
https://niehs.github.io/beethoven/
Other
2 stars 0 forks source link

Applying `brulee` for MLP #300

Open sigmafelix opened 3 months ago

sigmafelix commented 3 months ago

https://github.com/tidymodels/brulee

@mitchellmanware It looks like converting input dataset into torch tensors then fit multilayer perceptron. Tidymodels can use this package as an engine for a workflow. I'm trying this package for our MLP and hopefully it will be compatible with other teams' work.

CNN application is on hold.

mitchellmanware commented 3 months ago

Wow, this is a very powerful package. This will make the building of the MLP much easier but also nice and simple for reproduction.

mitchellmanware commented 1 month ago

https://github.com/mitchellmanware/scratch/blob/main/beethoven/rtorch/brulee_function.R

sigmafelix commented 1 month ago

@mitchellmanware Thank you for sharing the great work! I learned that one of branches brulee maintainers have named gpu enabled GPU capabilities for brulee functions. I am struggling with connecting CUDA drivers on DDN to the interactive session, hopefully I could share more about that then gain computational efficiency with GPU.