thunlp / OpenNE

An Open-Source Package for Network Embedding (NE)
MIT License
1.68k stars 485 forks source link

Current requirements allow incompatible versions of libraries to be installed #109

Closed apcomello closed 3 years ago

apcomello commented 4 years ago

When I try to install the requirements using the requirements.txt file, per the installation instructions, I end up with incompatible versions of the required libraries. These are the two specific issues I had:

I've tested the framework with newer versions of numpy, so it seems possible to use the newest version for tensorflow 1 as long as the numpy version is updated accordingly, but at the very least there is a need to specify that you cannot use tensorflow 2 with this.

droully commented 4 years ago

i recommend you to use the PyTorch branch: https://github.com/thunlp/OpenNE/tree/pytorch