conflict numpy version between tensorflow and tf-nighty

[hoaquocphan]

When I install tf-nighty as below command:
pip3 install tf-nightly

the error appear as below:

Requirement already satisfied: zipp>=0.5 in /home/hoaphan/.local/lib/python3.6/site-packages (from importlib-metadata->markdown>=2.6.8->tb-nightly~=2.4.0.a->tf-nightly) (3.4.0)
Installing collected packages: numpy
  Attempting uninstall: numpy
    Found existing installation: numpy 1.18.5
    Uninstalling numpy-1.18.5:
      Successfully uninstalled numpy-1.18.5
ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.
tensorflow 2.3.1 requires numpy<1.19.0,>=1.16.0, but you have numpy 1.19.5 which is incompatible.
Successfully installed numpy-1.19.5

this mean:
the tensorflow requires numpy <1.19.0
but the tf-nighty requires numpy 1.19.5

so I cannot install tensorflow and tf-nighty at same time,
so please help me solve this issue.

[amahendrakar]

@hoaquocphan,
TF v2.3.2 requires numpy >= 1.16.0, < 1.19.0, whereas for TF v2.4 and the latest TF-nightly the required version is numpy~=1.19.2.

In this case, I'd suggest you to install numpy v1.19.2, as it is compatible with the latest stable version of TensorFlow i.e. v2.4.1.

so I cannot install tensorflow and tf-nighty at same time,

Also, you can only import any one release of TensorFlow in the script. So, I'd recommend using separate virtual environments for each version.

Thanks!

[mihaimaruseac]

Since tf-nightly and tensorflow use the same code, you should never install both in the same environment. Usually, the latest one overrides most of the other. This is the way pip works and TF can do nothing in this case. That's why you should use venv/virtualenv.

[google-ml-butler]

Are you satisfied with the resolution of your issue?
Yes
No