Each such requirements.txt file contains the list of (additional) packages that must be installed prior to running the corresponding .py codes.
I am unable to install the these files.
I tried installing them both on a cluster (running Red Hat) and on a Ubuntu virtual machine installed on the cluster.
I tried installing them using pip (i.e. using
Code: Select all
pip install -r requirements.txt
Code: Select all
conda install --file requirements.txt
Irrespective of the OS used, I always get the same errors.
When trying to install the file via pip, the error is:
Code: Select all
ERROR: Invalid requirement: 'python-scipy = 0.18.1-2~pn1' (from line 13 of dependencies.txt)
Hint: = is not a valid operator. Did you mean == ?
Code: Select all
InvalidVersionSpec: Invalid version '0.18.1-2~pn1': invalid character(s)
Code: Select all
python-tensorflow = 1.0.0-0~pn1
python-numpy = 1:1.12.0-2~pn0
the prefix "python-" before different packages (e.g. Numpy, Tensorflow)
the suffix "~pn0" or "~pn1" and the end.
I think all the requirements files I have were actually generated on a Debian OS, but I am not sure how to confirm this and I am not sure if there is any method that would allow me the handle these file (i.e. simply install them, via pip, conda or any other way). My questions are the following:
1. Is there any way to decide if the requirements.txt files that contain the elements listed above were generated on a Debian OS?
2. Could anyone suggest a way to install such files on a RedHat or Ubuntu OS? Are there other options beside pip and conda that I should try?