Skip to content
This repository has been archived by the owner on Jun 9, 2021. It is now read-only.

The new Apple M1 MLcompute Tensorflow2.4 not compatible with Numpy 1.20.1, after attempting installing ~>1.19.2, got error #220

Open
xxl4tomxu98 opened this issue Apr 2, 2021 · 5 comments

Comments

@xxl4tomxu98
Copy link

OS Platform and Distribution (e.g., Linux Ubuntu 16.04): Mac Air M1
Mobile device (e.g. iPhone 8, Pixel 2, Samsung Galaxy) if the issue happens on mobile device:
TensorFlow installed from (source or binary): Github Apple MLcompute repo
TensorFlow version: 2.4
Python version: 3.8.6
Installed using virtualenv? pip? conda?: conda, miniforge3

After #48237, I uninstalled 1.20.1 and 1.18.5 and tried to install 1.19.2 or >1.19.2. error. I uninstalled numpy 1.20.1 and 1.18.5 and tried to install numpy=1.19.2 and got this error:

ERROR: Failed building wheel for numpy
Failed to build numpy
ERROR: Could not build wheels for numpy which use PEP 517 and cannot be installed directly.

The original error was that non compatible issue between tf2.4 and np1.20.1.

@calvalugo
Copy link

The SciPy Ecosystem is currently unsupported on the M1 Macs.

@xxl4tomxu98
Copy link
Author

@calvalugo However, my issue appears to be related to numpy and tensorflow compatibility, not spicy.

@c-juhwan
Copy link

Are you trying to install numpy with pip?
I've also seen this error when I tried to install with pip.

I tried to use conda and everything's now fine.

conda install numpy=1.19.5

@xxl4tomxu98
Copy link
Author

This part is also confusing. The apple mlcompute package of tf2.4 supposedly already has numpy included. I don’t know what version that is? I tried condo install of numpy and did not work

@ManuelSchneid3r
Copy link

ManuelSchneid3r commented Apr 29, 2021

@xxl4tomxu98 Check #153. And please close if resolved.

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

No branches or pull requests

4 participants