(solved) Python can not import keras

[deepakdeshp]

Hello,
I installed keras yet I cant import it. What am I missing?

Code: Select all

 pip3 show keras
Name: Keras
Version: 2.4.3
Summary: Deep Learning for humans
Home-page: https://github.com/keras-team/keras
Author: Francois Chollet
Author-email: francois.chollet@gmail.com
License: MIT
Location: /home/uma/anaconda3/lib/python3.7/site-packages
Requires: h5py, scipy, numpy, pyyaml
Required-by: 
(base) uma@uma ~ $ python3.9
Python 3.9.5 (default, May 19 2021, 11:32:47) 
[GCC 9.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import tensorflow
>>> import keras
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ModuleNotFoundError: No module named 'keras'  

[spamegg]

Code: Select all

import numpy as np
import tensorflow as tf
from tensorflow import keras

https://keras.io/getting_started/intro_ ... engineers/

[deepakdeshp]

Code: Select all

  from tensorflow import keras
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ImportError: cannot import name 'keras' from 'tensorflow' (unknown location)
 

[deepakdeshp]

Code: Select all

pip3 show numpy
Name: numpy
Version: 1.20.3
Summary: NumPy is the fundamental package for array computing with Python.
Home-page: https://www.numpy.org
Author: Travis E. Oliphant et al.
Author-email: None
License: BSD
Location: /home/uma/anaconda3/lib/python3.7/site-packages
Requires: 
Required-by: tensorflow, tensorboard, tables, statsmodels, seaborn, scipy, scikit-learn, PyWavelets, pytest-arraydiff, patsy, pandas, opt-einsum, opencv-python, numexpr, numba, mkl-random, mkl-fft, matplotlib, Keras, Keras-Preprocessing, imageio, h5py, Bottleneck, bokeh, bkcharts, astropy
(base) uma@uma ~ $ python3.9
Python 3.9.5 (default, May 19 2021, 11:32:47) 
[GCC 9.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import numpy
Traceback (most recent call last):
  File "/usr/lib/python3/dist-packages/numpy/core/__init__.py", line 17, in <module>
    from . import multiarray
  File "/usr/lib/python3/dist-packages/numpy/core/multiarray.py", line 14, in <module>
    from . import overrides
  File "/usr/lib/python3/dist-packages/numpy/core/overrides.py", line 7, in <module>
    from numpy.core._multiarray_umath import (
ModuleNotFoundError: No module named 'numpy.core._multiarray_umath'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/lib/python3/dist-packages/numpy/__init__.py", line 142, in <module>
    from . import core
  File "/usr/lib/python3/dist-packages/numpy/core/__init__.py", line 47, in <module>
    raise ImportError(msg)
ImportError: 

IMPORTANT: PLEASE READ THIS FOR ADVICE ON HOW TO SOLVE THIS ISSUE!

Importing the numpy c-extensions failed.
- Try uninstalling and reinstalling numpy.
- If you have already done that, then:
  1. Check that you expected to use Python3.9 from "/usr/bin/python3.9",
     and that you have no directories in your PATH or PYTHONPATH that can
     interfere with the Python and numpy version "1.17.4" you're trying to use.
  2. If (1) looks fine, you can open a new issue at
     https://github.com/numpy/numpy/issues.  Please include details on:
     - how you installed Python
     - how you installed numpy
     - your operating system
     - whether or not you have multiple versions of Python installed
     - if you built from source, your compiler versions and ideally a build log

- If you're working with a numpy git repository, try `git clean -xdf`
  (removes all files not under version control) and rebuild numpy.

Note: this error has many possible causes, so please don't comment on
an existing issue about this - open a new one instead.

Original error was: No module named 'numpy.core._multiarray_umath'

[spamegg]

You are using Python 3.9 but the pip3 packages might have been installed for Python 3.8 (system) so there might be a confusion there. Not sure.

I would recommend removing Python 3.9 completely, and installing numpy, tensorflow, keras for the system Python.

If you want different Python versions then use pyenv https://github.com/pyenv/pyenv/

[deepakdeshp]

You are using Python 3.9 but the pip3 packages might have been installed for Python 3.8 (system) so there might be a confusion there. Not sure.

I would recommend removing Python 3.9 completely, and installing numpy, tensorflow, keras for the system Python.

If you want different Python versions then use pyenv https://github.com/pyenv/pyenv/

Thanks. I am able to import keras and numpy with python3.7. But I am not sure how to set the GPU to use tensorflow.

Code: Select all

  python3.7
Python 3.7.4 (default, Aug 13 2019, 20:35:49) 
[GCC 7.3.0] :: Anaconda, Inc. on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import tensorflow
2021-06-16 21:34:12.608970: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libcudart.so.11.0'; dlerror: libcudart.so.11.0: cannot open shared object file: No such file or directory
2021-06-16 21:34:12.609032: I tensorflow/stream_executor/cuda/cudart_stub.cc:29] Ignore above cudart dlerror if you do not have a GPU set up on your machine.
>>> from tensorflow import keras
>>> import numpy
>>> 
KeyboardInterrupt
>>> 
(base) uma@uma ~ $ inxi -G
Graphics:
  Device-1: AMD Picasso driver: amdgpu v: kernel 
  Display: x11 server: X.Org 1.20.9 driver: modesetting unloaded: fbdev,vesa 
  resolution: 1920x1080~60Hz 
  OpenGL: renderer: AMD RAVEN (DRM 3.40.0 5.11.0-20-generic LLVM 10.0.0) 
  v: 4.6 Mesa 20.0.0-devel - padoka PPA 

[deepakdeshp]

Code: Select all

python3.7
Python 3.7.4 (default, Aug 13 2019, 20:35:49) 
[GCC 7.3.0] :: Anaconda, Inc. on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import tensorflow as tf
2021-06-16 21:39:56.561764: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libcudart.so.11.0'; dlerror: libcudart.so.11.0: cannot open shared object file: No such file or directory
2021-06-16 21:39:56.561816: I tensorflow/stream_executor/cuda/cudart_stub.cc:29] Ignore above cudart dlerror if you do not have a GPU set up on your machine.
>>> print("Num GPUs Available: ", len(tf.config.list_physical_devices('GPU')))
2021-06-16 21:40:02.817477: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libcuda.so.1'; dlerror: libcuda.so.1: cannot open shared object file: No such file or directory
2021-06-16 21:40:02.817550: W tensorflow/stream_executor/cuda/cuda_driver.cc:326] failed call to cuInit: UNKNOWN ERROR (303)
2021-06-16 21:40:02.817594: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:156] kernel driver does not appear to be running on this host (uma): /proc/driver/nvidia/version does not exist
Num GPUs Available:  0
  

[spamegg]

You have AMD Graphics, I think you need an Nvidia graphics card for that.

Code: Select all

2021-06-16 21:40:02.817550: W tensorflow/stream_executor/cuda/cuda_driver.cc:326] failed call to cuInit: UNKNOWN ERROR (303)
2021-06-16 21:40:02.817594: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:156] kernel driver does not appear to be running on this host (uma): /proc/driver/nvidia/version does not exist

CUDA is Nvidia's own GPU computation framework.

Unfortunately most machine learning/GPU-compute libraries do not support AMD. You might try PlaidML: https://github.com/plaidml/plaidml

[deepakdeshp]

Is there a way to install tensorflow and numpy for python3.9?

[spamegg]

Is there a way to install tensorflow and numpy for python3.9?

Use pyenv. Won't make a difference though, you still don't have Nvidia.

[deepakdeshp]

Is there a way to install tensorflow and numpy for python3.9?

Use pyenv. Won't make a difference though, you still don't have Nvidia.

I am not talking about the display here. Now tensorflow and keras runs under python3.7 and gives error in python3.9 Is there a way to run them under python3.9?

[spamegg]

Yes, with pyenv it should be possible. You remove your 3.9 (which I assume you got from deadsnakes-ppa or built from source manually), install pyenv, then install 3.9 using pyenv, then activate that 3.9 using pyenv, then install those packages again, which will get installed for that 3.9 inside pyenv's virtual environment separate from the system Python. It will be installed to a folder like ~/.pyenv/versions/3.9.5/lib/python3.9

[deepakdeshp]

Thank you spamegg for your patience.

[tkocou]

For someone wanting to use AMD GPUs with tensorflow, this article may help:
https://medium.com/nerd-for-tech/use-yo ... 7e34ff3554