Solution without upgrading numpy

While upgrading the numpy version would often solve the issue, it's not always viable. Good example is the case when you're using tensorflow==2.6.0 which isn't compatible with the newest numpy version (it requires ~=1.19.2).

As already mentioned in FZeiser's answer, there was a change in numpys C API in version 1.20.0. There are packages that rely on this C API when they are being built, e.g. pycocotools. Given that pips dependency resolver doesn't guarantee any order for installing the packages, the following might happen:

1. pip figures out that it needs to install numpy and it chooses the latest version, 1.21.2 as of the time writing this answer.
2. It then builds a package that depends on numpy and its C API, e.g. pycocotools. This package is now compatible with numpy 1.21.2 C API.
3. At a later point pip needs to install a package that has a requirement for an older version of numpy, e.g. tensorflow==2.6.0 which would try to install numpy==1.19.5. As a result, numpy==1.21.2 is uninstalled and the older version is installed.
4. When running code that uses pycocotools, its current installation relies on the updated numpy C API, yet the numpy version was downgraded which would result in the error.

Solution

You should rebuild the package with the outdated numpy C API usage, e.g. for pycocotools:

pip uninstall pycocotools
pip install pycocotools --no-binary pycocotools

numpy version 1.22 solved it for me.

I was facing same issue in raspberry pi 3. Actually the error is with pandas. Although tensorflow need numpy~=1.19.2 , but pandas is not compliable with it. So, I have upgraded (because downgrading is not) my numpy to latest version and all works fine!!!!.

root@raspberrypi:/home/pi# python3
Python 3.7.3 (default, Jan 22 2021, 20:04:44) 
[GCC 8.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import numpy as np 
>>> np.__version__
'1.21.5'
>>> import pandas as pd
>>> pd.__version__
'1.3.5'
>>> import tensorflow as tf
>>> tf.__version__
'2.4.0'
>>> tf.keras.__version__
'2.4.0'
>>> tf.keras.layers
<module 'tensorflow.keras.layers' from '/usr/local/lib/python3.7/dist-packages/tensorflow/keras/layers/__init__.py'>

Same issue here also - https://github.com/bitsy-ai/tensorflow-arm-bin/issues/5

Tensorflow source: https://github.com/bitsy-ai/tensorflow-arm-bin

Install older version of gensim, It works!

pip install gensim==3.5.0

or

conda install gensim==3.5.0

Use python virtual environments and install gensim using :

pip install gensim==3.8.3

Upgrade numpy version:

pip install -U numpy

Indeed, (building and) installing with numpy>=1.20.0 should work, as pointed out e.g. by this answer below. However, I thought some background might be interesting -- and provide also alternative solutions.

There was a change in the C API in numpy 1.20.0. In some cases, pip seems to download the latest version of numpy for the build stage, but then the program is run with the installed version of numpy. If the build version used in <1.20, but the installed version is =>1.20, this will lead to an error.

(The other way around it should not matter, because of backwards compatibility. But if one uses an installed version numpy<1.20, they did not anticipate the upcoming change.)

This leads to several possible ways to solve the problem:

• upgrade (the build version) to numpy>=1.20.0
• use minmum supported numpy version in pyproject.toml (oldest-supported-numpy)
• install with --no-binary
• install with --no-build-isolation

For a more detailed discussion of potential solutions, see https://github.com/scikit-learn-contrib/hdbscan/issues/457#issuecomment-773671043.

For anyone using Poetry it is necessary to have experimental.new-installer set to true for an application with a numpy<1.20 dependency to be built correctly i.e:

poetry config experimental.new-installer true

It is true by default but if (as was the case for me) it has been changed it can catch you out.

My application uses Tensorflow and I did not therefore have the option of upgrading to >1.20. Poetry also does not support --no-binary dependencies.

After you pip install any package, makes sure you restart the Kernel and should work. usually packages get upgraded automatically and all you need is a quick restart. At least, this what worked in my situation and I was getting the same error when I tried to install and use pomegranate.

what worked for me was:

pip uninstall numpy
conda install -y -c conda-forge numpy

as bizzare as it might sound...I didn't even have to uninstall it with conda which seemed odd to me. I am using python 3.9

Upgrading python version to numpy ==1.21.1 worked for me!

I'm in Python 3.8.5. It sounds too simple to be real, but I had this same issue and all I did was reinstall numpy. Gone.

pip install --upgrade numpy

or

pip uninstall numpy
pip install numpy

I had this issue when using the tensorflow object api. Tensorflow is currently NOT compatible with numpy==1.20 (although this issue is not apparent until later). In my case, the issue was caused by pycocotools. I fixed by installing an older version.

pip install pycocotools==2.0.0

try with numpy==1.20.0 this worked here, even though other circumstances are different (python3.8 on alpine 3.12).

For almost the same image : python:3.7-slim-buster

I started to have this problem just today, it was non exitent before.

I solved it by removing numpy from requirement.txt file and doing instead the following in my Dockerfile:

RUN pip3 install --upgrade  --no-binary numpy==1.18.1 numpy==1.18.1 \
&& pip3 install -r requirements.txt 

I use some old versions of keras and its librairies and updrading to numpy 1.20.0 didn't work for those librairies. But I think the solution consist in the first command that I gave you wich tell pip to try to not compile numpy and download a pre-compiled version.

The trick in the command is that you might find people telling you to use --no-binary option of pip to solve the problem, but they don't specify how and it might be tricky (as it happened to me); you have to write the package two times in the command in order for it to work or else pip will throw you an error.

I think the --upgrade option in the first command isn't necessary.