Unable to install due to numpy==1.19.4

[thompsonalecbgo]

Hi I am unable to use gradient on python 3.7.9 due to RuntimeError caused by numpy==1.19.4

RuntimeError: The current Numpy installation ('c:\users\XXXX\envs\gradient\lib\site-packages\numpy\init.py') fails to pass a sanity check due to a bug in the windows runtime. See this issue for more information: https://tinyurl.com/y3dm3h86

I tried downgrading the numpy, but it appears to be incompatible and i got the ff ModuleNotFoundError:
ModuleNotFoundError: No module named '_curses'

[lborke]

I have had this issue a few days before.

See also https://developercommunity.visualstudio.com/content/problem/1207405/fmod-after-an-update-to-windows-2004-is-causing-a.html

The best current solution under Win (Version 2004) is:

pip uninstall numpy

then

pip install numpy==1.19.3

Best

[lborke]

PS:
I am not using numpy together with gradient-cli.
My suggestion addresses only the numpy issue.

One solution for gradient-cli (+ numpy) would be using Docker/Hyper-V under Windows to have a native Linux system...

[sanfilip]

@weikaolun @lamroger I am still able to repo this issue with the latest gradient-cli which pulls in gradient-utils-0.3.1
tensoflow-2.3.1 which is the latest, is not compatible due to the numpy conflict. Seems like we should downgrade gradient-utils to numpy 1.18.5 at least. Also note there is a conflict with hyperopt

Steps to repro:

on macos:
which python3
/Library/Frameworks/Python.framework/Versions/3.7/bin/python3
python3 --version
Python 3.7.3
python3 -m venv venv
source venv/bin/activate
pip install gradient
pip install tensorflow
...
(error:)
gradient-utils 0.3.1 has requirement numpy==1.19.3, but you'll have numpy 1.18.5 which is incompatible.

Also gradient-sdk is no longer compatible with gradient-utils; we need to revise samples to account for this.

[sanfilip]

@weikaolun @lamroger
This fails for any version of tensorflow, 1.15, 2.x

[Paulj1989]

I've had similar issues when trying to install gradient. I'm trying to install in a pipenv, and it produces a resolution failure when resolving dependencies.

The error it returns is as follows:

ERROR: Could not find a version that matches numpy==1.18.5,==1.19.3,>=1.12.0,>=1.13.0,>=1.13.3,>=1.14.5,>=1.15,>=1.15.4,>=1.7,>=1.9.1,~=1.19.2

And the incompatible versions:

There are incompatible versions in the resolved dependencies:
  numpy==1.19.3 (from -r /tmp/pipenvn4iqexjvrequirements/pipenv-5gja4g8i-constraints.txt (line 6))
  numpy==1.18.5 (from gradient-utils==0.3.2->gradient==1.3.0->-r /tmp/pipenvn4iqexjvrequirements/pipenv-5gja4g8i-constraints.txt (line 10))
  numpy>=1.12.0 (from tensorboard==2.4.0->tensorflow==2.4.0->-r /tmp/pipenvn4iqexjvrequirements/pipenv-5gja4g8i-constraints.txt (line 15))
  numpy>=1.12.0 (from tensorflow-hub==0.10.0->-r /tmp/pipenvn4iqexjvrequirements/pipenv-5gja4g8i-constraints.txt (line 2))
  numpy>=1.13.0 (from yellowbrick==1.2->-r /tmp/pipenvn4iqexjvrequirements/pipenv-5gja4g8i-constraints.txt (line 7))
  numpy>=1.13.3 (from scikit-learn==0.23.2->sklearn==0.0->-r /tmp/pipenvn4iqexjvrequirements/pipenv-5gja4g8i-constraints.txt (line 13))
  numpy>=1.14.5 (from scipy==1.5.4->keras==2.4.3->-r /tmp/pipenvn4iqexjvrequirements/pipenv-5gja4g8i-constraints.txt (line 12))
  numpy>=1.15 (from matplotlib==3.3.3->-r /tmp/pipenvn4iqexjvrequirements/pipenv-5gja4g8i-constraints.txt (line 4))
  numpy>=1.15 (from seaborn==0.11.0->-r /tmp/pipenvn4iqexjvrequirements/pipenv-5gja4g8i-constraints.txt (line 11))
  numpy>=1.15.4 (from pandas==1.1.5->-r /tmp/pipenvn4iqexjvrequirements/pipenv-5gja4g8i-constraints.txt (line 14))
  numpy>=1.7 (from h5py==2.10.0->keras==2.4.3->-r /tmp/pipenvn4iqexjvrequirements/pipenv-5gja4g8i-constraints.txt (line 12))
  numpy>=1.7 (from opt-einsum==3.3.0->tensorflow==2.4.0->-r /tmp/pipenvn4iqexjvrequirements/pipenv-5gja4g8i-constraints.txt (line 15))
  numpy>=1.9.1 (from keras-preprocessing==1.1.2->tensorflow==2.4.0->-r /tmp/pipenvn4iqexjvrequirements/pipenv-5gja4g8i-constraints.txt (line 15))
  numpy>=1.9.1 (from keras==2.4.3->-r /tmp/pipenvn4iqexjvrequirements/pipenv-5gja4g8i-constraints.txt (line 12))
  numpy~=1.19.2 (from tensorflow==2.4.0->-r /tmp/pipenvn4iqexjvrequirements/pipenv-5gja4g8i-constraints.txt (line 15))

Has anyone found a resolution?

[lamroger]

I've had similar issues when trying to install gradient. I'm trying to install in a pipenv, and it produces a resolution failure when resolving dependencies.

The error it returns is as follows:

ERROR: Could not find a version that matches numpy==1.18.5,==1.19.3,>=1.12.0,>=1.13.0,>=1.13.3,>=1.14.5,>=1.15,>=1.15.4,>=1.7,>=1.9.1,~=1.19.2

And the incompatible versions:

There are incompatible versions in the resolved dependencies:
  numpy==1.19.3 (from -r /tmp/pipenvn4iqexjvrequirements/pipenv-5gja4g8i-constraints.txt (line 6))
  numpy==1.18.5 (from gradient-utils==0.3.2->gradient==1.3.0->-r /tmp/pipenvn4iqexjvrequirements/pipenv-5gja4g8i-constraints.txt (line 10))
  numpy>=1.12.0 (from tensorboard==2.4.0->tensorflow==2.4.0->-r /tmp/pipenvn4iqexjvrequirements/pipenv-5gja4g8i-constraints.txt (line 15))
  numpy>=1.12.0 (from tensorflow-hub==0.10.0->-r /tmp/pipenvn4iqexjvrequirements/pipenv-5gja4g8i-constraints.txt (line 2))
  numpy>=1.13.0 (from yellowbrick==1.2->-r /tmp/pipenvn4iqexjvrequirements/pipenv-5gja4g8i-constraints.txt (line 7))
  numpy>=1.13.3 (from scikit-learn==0.23.2->sklearn==0.0->-r /tmp/pipenvn4iqexjvrequirements/pipenv-5gja4g8i-constraints.txt (line 13))
  numpy>=1.14.5 (from scipy==1.5.4->keras==2.4.3->-r /tmp/pipenvn4iqexjvrequirements/pipenv-5gja4g8i-constraints.txt (line 12))
  numpy>=1.15 (from matplotlib==3.3.3->-r /tmp/pipenvn4iqexjvrequirements/pipenv-5gja4g8i-constraints.txt (line 4))
  numpy>=1.15 (from seaborn==0.11.0->-r /tmp/pipenvn4iqexjvrequirements/pipenv-5gja4g8i-constraints.txt (line 11))
  numpy>=1.15.4 (from pandas==1.1.5->-r /tmp/pipenvn4iqexjvrequirements/pipenv-5gja4g8i-constraints.txt (line 14))
  numpy>=1.7 (from h5py==2.10.0->keras==2.4.3->-r /tmp/pipenvn4iqexjvrequirements/pipenv-5gja4g8i-constraints.txt (line 12))
  numpy>=1.7 (from opt-einsum==3.3.0->tensorflow==2.4.0->-r /tmp/pipenvn4iqexjvrequirements/pipenv-5gja4g8i-constraints.txt (line 15))
  numpy>=1.9.1 (from keras-preprocessing==1.1.2->tensorflow==2.4.0->-r /tmp/pipenvn4iqexjvrequirements/pipenv-5gja4g8i-constraints.txt (line 15))
  numpy>=1.9.1 (from keras==2.4.3->-r /tmp/pipenvn4iqexjvrequirements/pipenv-5gja4g8i-constraints.txt (line 12))
  numpy~=1.19.2 (from tensorflow==2.4.0->-r /tmp/pipenvn4iqexjvrequirements/pipenv-5gja4g8i-constraints.txt (line 15))

Has anyone found a resolution?

Hi @Paulj1989, would you be able to use tensorflow 2.3.1 instead? We pinned numpy to 1.18.5 in gradient-utils to resolve the above issues and tensorflow 2.3.1 should be compatible.

[Paulj1989]

Hi @Paulj1989, would you be able to use tensorflow 2.3.1 instead? We pinned numpy to 1.18.5 in gradient-utils to resolve the above issues and tensorflow 2.3.1 should be compatible.

Success! Thanks for your help.