Skip to content

Instantly share code, notes, and snippets.

@IAmSuyogJadhav

IAmSuyogJadhav/install_tsne-cuda.sh

Created January 22, 2020 21:52
Show Gist options
  • Star 0 You must be signed in to star a gist
  • Fork 0 You must be signed in to fork a gist
  • Save IAmSuyogJadhav/ceb57ef3b662bab2788fde269f3c864f to your computer and use it in GitHub Desktop.
Save IAmSuyogJadhav/ceb57ef3b662bab2788fde269f3c864f to your computer and use it in GitHub Desktop.
Download ZIP
Script to install tsne-CUDA by @CannyLab using conda. Derived from https://medium.com/analytics-vidhya/super-fast-tsne-cuda-on-kaggle-b66dcdc4a5a4 to work on any linux system
Raw
install_tsne-cuda.sh
# Change CUDA version if needed
conda install faiss-gpu cudatoolkit=10.0 -c pytorch
sudo apt install libopenblas-dev
wget https://anaconda.org/CannyLab/tsnecuda/2.1.0/download/linux-64/tsnecuda-2.1.0-cuda100.tar.bz2
tar xvjf tsnecuda-2.1.0-cuda100.tar.bz2 --wildcards 'lib/*'
tar xvjf tsnecuda-2.1.0-cuda100.tar.bz2 --wildcards 'site-packages/*'
# Fill your username below. Change python version if needed
cp -r site-packages/* /home/username/anaconda3/lib/python3.7/site-packages/
sudo mkdir -p /usr/local/cuda/lib64
cp ./lib/libfaiss.so /usr/local/cuda/lib64/
# To force update the library path
sudo echo /usr/local/cuda/lib64/ >> /etc/ld.so.conf
sudo ldconfig
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment