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
4 stars 0 forks source link

Base Learner Teams #298

Open kyle-messier opened 6 months ago

kyle-messier commented 6 months ago
  1. xgboost: @eva0marques @Sanisha003
  2. [C]NN: @sigmafelix @sigmafelix
  3. prestoGP: @dzilber @dawranadeep
sigmafelix commented 3 months ago

GPU settings

  1. XGBoost

    • Setting up a user library path where you want (in my case, ~/r-libs)
    • Download prebuilt GPU version library at https://github.com/dmlc/xgboost/releases/tag/v2.0.3
    • Run command R CMD INSTALL [downloaded tarball file path] --library=~/r-libs
    • Bind CUDA (please refer to the post on Teams)

  2. Brulee

    • Bind CUDA
    • Install PyTorch (*note -- R torch currently supports CUDA 11.8 is supported on HPC) 
      pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
    • In R, run pak::pak("tidymodels/brulee@gpu")