The module python-ml has been updated.

The module "python-ml" is a unified python module for machine learning at the department of mathematics, superseding the specific modules for tensorflow, pytorch, and other libraries.

The following two modules are being moved into production:

  •  python-ml/3.7.3-2019.05.22-cpu
  •  python-ml/3.7.3-2019.05.22-cuda

These modules include the following python packages:

  •  tensorflow 1.14.0-rc1
  •  pytorch 1.1.0
  •  torchvision 0.3.0
  •  sklearn 0.21.2
  •  keras 2.2.4

They are all built against CUDA10 (10.1 where possible), and should work[tm]. Please test the modules a bit and see if they can be used as a path forward for python-based machine learning at the department. Also, for those of you who like R, please also note that these are the backend packages for the module R/3.5.3-ml-gcc with keras and tensorflow support.

 

Published June 11, 2019 9:36 PM - Last modified June 11, 2019 9:36 PM