Dependency issues trying to install required packages

[vrooje]

I'm using conda to try and install the required Python version and package versions, but I cannot resolve a conflict between the numpy and tensorflow versions that are required. I've tried stepping up the version of numpy to resolve it (in hopes this won't break the code), but resolving it requires upgrading to Python 3.6, and I've been getting errors.

Can you provide an installable environment or some advice on getting through this?

Many thanks.

[mrciolino]

I have had good success with ... Maybe try in a new conda enviroment.

conda install keras
conda install tqdm
conda install pillow
conda install scikit-image
conda install scipy==1.1.0

scipy was the only package that I needed to downgrade scipy for. In terms of an installable environment, docker is prob what your looking for. They have some good premade environments on dockerhub that you can "docker pull" and get up an running...
https://hub.docker.com/r/ufoym/deepo