I can relate to your struggle and would be glad to offer some advice. Since I do not have insight into your background with environment management in
python, I'll add some basic details - apologies for some things that might appear obvious.
Fluxy is completely right - you need to get these proxy servers. When I first started in a company that had similar restrictions (and
zero IT support for programming), I had to reach out to other
python/R programmers to get the proxy servers. Make sure to get both http and https if both are available. Perhaps IT team can help you with this too, but it might take you forever to find the right person in IT department.
Once you obtain proxy servers:
We default to managing environments through conda - and since you already have this installed, try this:
1) Start Anaconda Prompt
2) type in command prompt (replacing "smth" with your proxy servers)
set HTTP_PROXY=http:/smth
set HTTPS_PROXY=https:/smth
3) conda install packageyouneed
If this doesn't work, try specifying the channel from which you want to install the package.
Long-term solution:
If you do have .condarc file already (typically saved in your %userprofile%), just add proxy servers there. You can add that through command prompt or manually. If you do not have .condarc file setup, I highly recommend that you look into creating this file to permanenly set proxy server/channel priorities, etc. Check these instructions out:
Using Anaconda behind a company proxy — Anaconda documentation
Last resort option (not recommended!):
You can download the package manually. Conditional your company doesn't restrict downloads and you have access to Github. (Typically you can submit the request through supervisor/IT department to give you permission to view and download from Github if you're a developer).
Deep Learning for humans. Contribute to keras-team/keras development by creating an account on GitHub.
github.com
But beware if you download it into your base (or your most frequently used) environment - there might be package conflicts. Also, you will run into trouble when trying to detect this package in your conda environment (through conda list command).
Perhaps try downloading this zip file and unpacking it into another location on your machine rather than to your conda or
python environment and then try importing this package as a module by appending the path via sys. This way you do not touch your base
python env and there is less risk you'll have to de-install and re-install everything anew.
Every organization is different, so please take this advice with a grain of salt. Corrections from other users are welcomed! I went a bit too detailed here, but hope it helps you or someone else in the future.
www.educative.io
