numpy 1.19.0 is incompatible with tensorflow 2.2.1
kafasin opened this issue · comments
Warning while piping requirements,
-- ERROR: tensorflow 2.2.1 has requirement numpy<1.19.0,>=1.16.0, but you'll have numpy 1.19.0 which is incompatible.
Warning while piping requirements,
-- ERROR: tensorflow 2.2.1 has requirement numpy<1.19.0,>=1.16.0, but you'll have numpy 1.19.0 which is incompatible.
Hello @kafasin.
Just try to change numpy version from 1.19.0 to 1.18.5 in requirements.txt file. I haven't tested yet but the install process works fine, once version 1.18.5 solve all requirements for the packages that depend on numpy.
For more information about release 1.19.0 (like compatibility issues, deprecation and new features) follow this link.
However as mentioned by @abhishekkrthakur in #20, the book used TensorFlow 1.15 (last release of version 1). Due to diffs from ts v1 to v2, compatibility issues may occur (although the use of Keras API (tf.keras) can abstract these differences). In any case, there is a guide for migration between versions (see this link), or you can do as mentioned on README.md in this repo, changing the follow packages versions on requirements.txt file:
- gast: from 0.3.3 to 0.2.2
- numpy: from 1.19.0 to 1.18.5 (as mentioned before)
- tensorboard: from 2.2.2 to 1.15.0
- tensorflow: from 2.2.1 to 1.15.0
- tensorflow-estimator: from 2.2.0 to 1.15.1
I hope that could help you.
Cheers.