2

I am wondering that why I cannot install the TensorFlow 2.0.0 (stable version until now in their official website even in PyPi)

I used to had the TensorFlow 2.0.0b1 version, but I was dealing with several warnings because my numpy version wasn't compatible with the TensorFlow. It was 1.17 and must be 1.16.4 (Reference). Hence, I downgrade it but TensorFlow didn't work anymore. So I decide to reinstall the TensorFlow latest stable version as mentioned above.

Here is I've done so far:

pip install -U tensorflow

It trow backs 1.14 through pip show tensorflow

pip install tensorflow

It trow backs 1.14 through pip show tensorflow

pip install tensorflow==2.0
pip install tensorflow==2.0.0
pip install 'tensorflow>=2'

They trow back the following error:

  Could not find a version that satisfies the requirement tensorflow==2.0.0 (from versions: 0.12.1, 1.0.0, 1.0.1, 1.1.0rc0, 1.1.0rc1, 1.1.0rc2, 1.1.0, 1.2.0rc0, 1.2.0rc1, 1.2.0rc2, 1.2.0, 1.2.1, 1.3.0rc0, 1.3.0rc1, 1.3.0rc2, 1.3.0, 1.4.0rc0, 1.4.0rc1, 1.4.0, 1.4.1, 1.5.0rc0, 1.5.0rc1, 1.5.0, 1.5.1, 1.6.0rc0, 1.6.0rc1, 1.6.0, 1.7.0rc0, 1.7.0rc1, 1.7.0, 1.7.1, 1.8.0rc0, 1.8.0rc1, 1.8.0, 1.9.0rc0, 1.9.0rc1, 1.9.0rc2, 1.9.0, 1.10.0rc0, 1.10.0rc1, 1.10.0, 1.10.1, 1.11.0rc0, 1.11.0rc1, 1.11.0rc2, 1.11.0, 1.12.0rc0, 1.12.0rc1, 1.12.0rc2, 1.12.0, 1.12.2, 1.12.3, 1.13.0rc0, 1.13.0rc1, 1.13.0rc2, 1.13.1, 1.13.2, 1.14.0rc0, 1.14.0rc1, 1.14.0, 2.0.0a0, 2.0.0b0, 2.0.0b1)
No matching distribution found for tensorflow==2.0.0

[NOTE]:

OS: Ubuntu 16.04
Python: 3.6.8
numpy: 1.16.4
pip3 version: 18.1

[UPDATE OCT 2020]:

Apparently, this problem has been resolved, because I'm able to install tensorflow 2.3.1 at the moment without any problem on Ubuntu 16.04 Xenial.

Benyamin Jafari
  • 27,880
  • 26
  • 135
  • 150
  • 2
    Ubuntu 16.04 doesn't support the `manylinux2010` tag and TF 2 dropped support for `manylinux1` (as the glibc version was too old). Run `python -c "from pip._internal.pep425tags import get_supported; print(get_supported())"` to verify that - `manylinux2010` is not listed. Update to 18.04 to get the `manylinux2010` support. – hoefling Dec 27 '19 at 11:45
  • @hoefling Yes, I checked it, that was not. Thanks for the response. So should I use `2.0.0b1` version in Ubuntu 16.04? – Benyamin Jafari Dec 27 '19 at 11:50
  • @hoefling And what about `pip install tf-nightly` instead? – Benyamin Jafari Dec 27 '19 at 11:58
  • 2
    `tf-nightly` is basically a wheel built from current master, so it doesn't support `manylinux1` either since the latest releases don't. With xenial, the last installable version is 2.0.0b1, unfortunately. Surely, you can build tensorflow from source or play with installing a more recent glibc from a PPA, but IMO the easier path to take is to upgrade to bionic (if you can). – hoefling Dec 27 '19 at 13:01

1 Answers1

2

Mentioning the solution in this section (even though it is mentioned in the Comments Section), for the benefit of the community.

Any of the below steps will resolve the issue:

  1. Using the Tensorflow Version, 2.0.0b1 for the Ubuntu Version 16.04 (Xenial Xerus) Or
  2. By upgrading the Ubuntu OS Version from 16.04 (Xenial Xerus) to 18.04 (Bionic Beaver), we can install the Latest Tensorflow Version.
  • Apparently, this problem has been resolved, because I'm able to install *tensorflow 2.3.1* at the moment without any problem on *Ubuntu 16.04 Xenial*. – Benyamin Jafari Oct 24 '20 at 08:52