PackagesNotFoundError on running command to setup a suitable conda environment could look like this for the GPU version of Keras

Hey all, I want to try out KNIME’s shiny Keras integration and followed https://www.knime.com/deeplearning (I have anaconda and cuda setup, latest versions).

However, running the recommended command to setup a suitable conda environment could look like this for the GPU version of Keras (conda create -y -n py35_knime python=3.5 pandas jedi keras-gpu=2.1.2) yields the following error:

Solving environment: failed

PackagesNotFoundError: The following packages are not available from current channels:

  - keras-gpu=2.1.2

Current channels:

  - https://repo.continuum.io/pkgs/main/osx-64
  - https://repo.continuum.io/pkgs/main/noarch
  - https://repo.continuum.io/pkgs/free/osx-64
  - https://repo.continuum.io/pkgs/free/noarch
  - https://repo.continuum.io/pkgs/r/osx-64
  - https://repo.continuum.io/pkgs/r/noarch
  - https://repo.continuum.io/pkgs/pro/osx-64
  - https://repo.continuum.io/pkgs/pro/noarch

Using conda v 4.4.11 on OSX High Sierra 10.13.3. Are the installation instructions up to date?

Hi @motin,

I think the problem is that there is not really GPU support of Keras (i.e. keras-gpu) for MacOSX via anaconda available anymore. However, you don’t need keras-gpu necessarily to use the KNIME Deep Learning nodes. If you want GPU support, there are some blog-posts how to set keras up for it for MacOSX. Tensorflow, however, dropped GPU support for MacOSX with Tensorflow 1.2 as far as I understood it.

Cheers,

Christian