We Keep Coding

The right way to install stable-baselines?

I am trying to install stable-baselines and run the first two lines from Getting Started section of the online manual but no option is working. I started with
pip install stable-baselines
Now when I run:
import gym
from stable_baselines.common.policies import MlpPolicy
I get
No module named 'tensorflow.contrib'
This apparently is because tensorflow version 2 doesn't have tensorflow.contrib. But version 2 was released in Sept 2019. Do I really have to use only tensorflow version 1?
What is the right way to install stable-baselines and run that simple example?
I tried
pip install stable-baselines3
in a virtual environment. This gives a different error:
In [2]: from stable_baselines.common.policies import MlpPolicy
---------------------------------------------------------------------------
ModuleNotFoundError Traceback (most recent call last)
Input In [2], in <module>
----> 1 from stable_baselines.common.policies import MlpPolicy
ModuleNotFoundError: No module named 'stable_baselines'
In [3]: from stable_baselines3.common.policies import MlpPolicy
---------------------------------------------------------------------------
ImportError Traceback (most recent call last)
Input In [3], in <module>
----> 1 from stable_baselines3.common.policies import MlpPolicy
ImportError: cannot import name 'MlpPolicy' from 'stable_baselines3.common.policies' (/home/raph/RL/stable-baselines/venv/lib/python3.8/site-packages/stable_baselines3/common/policies.py)

To quote the github readme:
Note: Stable-Baselines supports Tensorflow versions from 1.8.0 to 1.14.0. Support for Tensorflow 2 API is planned.
The same github readme also recommends to use stable-baselines3, as stable-baselines is currently only being maintained and its functionality is not extended. Thus, I would not expect the TF1 -> TF2 update any time soon.
If you can not install this version of tensorflow, I suggest to use stable-baselines3 and follow the examples. The code you posted above is not consistent with the stable-baselines3 docs, which import the MlpPolicy as
from stable_baselines3.sac.policies import MlpPolicy

According to the stable-baselines documentation you can only use Tensorflow version 1.8.0 to version 1.15.0.
If you want to run Tensorflow 1, and you want to use pip as your package manager, you'll need to install python version 3.7 or lower. I did the following in my ubuntu terminal
sudo add-apt-repository ppa:deadsnakes/ppa
sudo apt update
sudo apt install python3.7
sudo apt install virtualenv
virtualenv --python=/usr/bin/python3.7 <env-name>
source <env-name>/bin/activate
pip install tensorflow==1.15.0
...
Alternatively, you could try using this guide, which gives instructions as to how to migrate something to Tensorflow version 2.

How to solve "AttributeError: module 'google.protobuf.descriptor' has no attribute '_internal_create_key"?

I encountered it while executing from object_detection.utils import label_map_util in jupyter notebook. It is actually the tensorflow object detection tutorial notebook(it comes with the tensorflow object detection api)
The complete error log:
AttributeError Traceback (most recent call last)
<ipython-input-7-7035655b948a> in <module>
1 from object_detection.utils import ops as utils_ops
----> 2 from object_detection.utils import label_map_util
3 from object_detection.utils import visualization_utils as vis_util
~\AppData\Roaming\Python\Python37\site-packages\object_detection\utils\label_map_util.py in <module>
25 import tensorflow as tf
26 from google.protobuf import text_format
---> 27 from object_detection.protos import string_int_label_map_pb2
28
29
~\AppData\Roaming\Python\Python37\site-packages\object_detection\protos\string_int_label_map_pb2.py in <module>
19 syntax='proto2',
20 serialized_options=None,
---> 21 create_key=_descriptor._internal_create_key,
22 serialized_pb=b'\n2object_detection/protos/string_int_label_map.proto\x12\x17object_detection.protos\"\xc0\x01\n\x15StringIntLabelMapItem\x12\x0c\n\x04name\x18\x01 \x01(\t\x12\n\n\x02id\x18\x02 \x01(\x05\x12\x14\n\x0c\x64isplay_name\x18\x03 \x01(\t\x12M\n\tkeypoints\x18\x04 \x03(\x0b\x32:.object_detection.protos.StringIntLabelMapItem.KeypointMap\x1a(\n\x0bKeypointMap\x12\n\n\x02id\x18\x01 \x01(\x05\x12\r\n\x05label\x18\x02 \x01(\t\"Q\n\x11StringIntLabelMap\x12<\n\x04item\x18\x01 \x03(\x0b\x32..object_detection.protos.StringIntLabelMapItem'
23 )
AttributeError: module 'google.protobuf.descriptor' has no attribute '_internal_create_key'

The protoc version I got through pip show protobuf and protoc --version were different. The version in pip was a bit outdated.
After I upgraded the pip version with
pip install --upgrade protobuf
the problem was solved.

These three commands solved it for me:
pip uninstall protobuf python3-protobuf
pip install --upgrade pip
pip install --upgrade protobuf

This can be due to incorrect versions of protobuf and python3-protobuf in your IDE(in my case PyCharm) and in site-packages. Following solved error in my case:
pip uninstall python3-protobuf
pip uninstall protobuf
then finally do
pip install protobuf
Also if you have requirements.txt or setup.py check protobuf version from pip freeze and use the same version in your requirements.txt file
pip freeze
(check protobuf version, and use same in requirements.txt)
protobuf==3.15.6 (in my case pip freeze gave me this version for protobuf)

For others looking into this more recently newer google libraries are using proto plus, a wrapper for python proto messages. Using this helper function worked for me (cred: tobked)
import json
import proto
def proto_message_to_dict(message: proto.Message) -> dict:
"""Helper method to parse protobuf message to dictionary."""
return json.loads(message.__class__.to_json(message))
https://github.com/googleapis/python-memcache/issues/19

After trying many different solutions (i'm working on a Mac) the one that worked for me is to reinstall protobuf using:
PROTOC_ZIP=protoc-3.7.1-osx-x86_64.zip
curl -OL https://github.com/protocolbuffers/protobuf/releases/download/v3.7.1/$PROTOC_ZIP
sudo unzip -o $PROTOC_ZIP -d /usr/local bin/protoc
sudo unzip -o $PROTOC_ZIP -d /usr/local 'include/*'
rm -f $PROTOC_ZIP
As is highlighted in this article

the problem for me was i had libprotobuf at 3.11 and protobuf at 3.15 upgrading libprotobuf fixed it for me.

I got this problem in Anaconda too.
I installed it with:
conda install -c conda-forge streamlit
with the versions protobuf and libprotobuf of 3.11.4
I was not able to update neither protobuf nor libprotobuf.
I recommend to install first "conda install protobuf" and then
conda install -c conda-forge streamlit
I got now the versions protobuf and libprotobuf of 3.14.0 and it works.

The issue of my condition is protoc not install.
step to solve the problem:
check pip3 show protobuf, you will see version of protobuf.
e.q. Version: 3.17.3
goto protobuf release page and install same version of protoc
e.q.
wget https://github.com/protocolbuffers/protobuf/releases/download/v3.17.3/protoc-3.17.3-linux-x86_64.zip
unzip protoc-3.17.3-linux-x86_64.zip
vim ~/.bashprofile and paste PATH=$PATH:/home/YOUR_HOST_NAME/bin
source ~/.bashprofile and echo $PATH to check whether /home/YOUR_HOST_NAME/bin exist or not.
check pip3 show protobuf is equal to protoc --version
If not work, reboot and try again.

This can also happen if you use protoc to generate your language specific bindings file in a different OS than where you import it. For example,
Linux OS
protoc -I=./ --python_out=./ my_module.proto
Mac OS
import my_module_pb2

Downgrade protobuf using pip
Sometimes the upgrade code pip install --upgrade protobuf throwed another error. -
version 3.20 seems to be working properly
pip install protobuf==3.20.*
similar to TypeError: Descriptors cannot not be created directly

Pytorch is installed but is not working on ubuntu 18.04

I am trying to install Pytorch via pip on ubuntu 18.04.I have python 3.6 and my laptop is HP-Pavilion notebook 15
The installation seems to be right because i get the message:
Installing collected packages: torch, torchvision Successfully
installed torch-1.3.1+cpu torchvision-0.4.2+cpu
i run the verification code and it is ok
from __future__ import print_function
import torch
x = torch.rand(5, 3)
print(x)
However, when i close the terminal or reboot and try to run he same code i get the error:
Traceback (most recent call last):
File "torch.py", line 2, in
import torch
AttributeError: module 'torch' has no attribute 'rand'

How are you executing the python script? Which python are you using? Maybe you installed the package in a different python version?
Try to set alias to the python you want to use:
alias python=/usr/local/bin/python3.6
Then pip install the package with that python alias you will always be using.
python pip install <package name>
Python now will install the package in the python files with the alias python - heading to files: /usr/local/bin/python3.6
Let me know if the error still occurs!

Install pytorch using pip through the below command:
pip3 install torch==1.3.1+cpu torchvision==0.4.2+cpu -f https://download.pytorch.org/whl/torch_stable.html
for any reference go through the official website of pytorch.

Change your file .py to another name, you named torch.py when you import torch it will call ur torch.py

Cannot import category_encoders module

I am not able to import category_encoders module in jupyter notebook in python 3 virtual environment.
Error
---------------------------------------------------------------------------
ModuleNotFoundError Traceback (most recent call last)
<ipython-input-15-86725efc8d1e> in <module>()
9 from plotly import graph_objs
10 from datetime import datetime
---> 11 import category_encoders as ce
12
13 import sklearn
ModuleNotFoundError: No module named 'category_encoders'
Output of "which pip"
/opt/virtual_env/py3/bin/pip
Output of "pip show category_encoders" is
Name: category-encoders
Version: 1.3.0
Summary: A collection sklearn transformers to encode categorical variables as numeric
Home-page: https://github.com/wdm0006/categorical_encoding
Author: Will McGinnis
Author-email: will#pedalwrencher.com
License: BSD
Location: /opt/virtual_env/py3/lib/python3.6/site-packages
Requires: numpy, pandas, statsmodels, scikit-learn, patsy, scipy
Required-by:
Output of "pip list"
Package Version
------------------- -------
absl-py 0.7.0
astor 0.7.1
backcall 0.1.0
category-encoders 1.3.0
cycler 0.10.0

pip install --upgrade category_encoders
Now,
Restart your kernel
import category_encoders as... so on and so forth
Also,
you can checkout the sk-learn's preprocessing class CategoricalEncoder...
from sklearn.preprocessing import CategoricalEncoder
get sk-learn from here,
pip install git+git://github.com/scikit-learn/scikit-learn.git
or Simply
pip install sklearn

I was using anaconda environment.
Initially, I installed category_encoders using pip (within the conda enviornment) and it failed to resolve. Then I uninstalled ``category_encoders` from pip
pip uninstall category-encoders
and installed it using conda:
conda install -c conda-forge category_encoders
That solved the issue.

If you have given your virtual environment ('py3' for your case) the permission for accessing the system site-packages directory you can try installing the package to your base (root) environment.
Search 'pyvenv.cfg' file in your system and check if it includes:
include-system-site-packages = true
If yes, install the package to your root environment as stated here and try importing it again in your Jupyter Notebook.

You can add a cell above and type ! pip install --upgrade category_encoders
It will install the package without opening a terminal window.

How to find which version of TensorFlow is installed in my system?

I need to find which version of TensorFlow I have installed. I'm using Ubuntu 16.04 Long Term Support.

This depends on how you installed TensorFlow. I am going to use the same headings used by TensorFlow's installation instructions to structure this answer.
Pip installation
Run:
python -c 'import tensorflow as tf; print(tf.__version__)' # for Python 2
python3 -c 'import tensorflow as tf; print(tf.__version__)' # for Python 3
Note that python is symlinked to /usr/bin/python3 in some Linux distributions, so use python instead of python3 in these cases.
pip list | grep tensorflow for Python 2 or pip3 list | grep tensorflow for Python 3 will also show the version of Tensorflow installed.
Virtualenv installation
Run:
python -c 'import tensorflow as tf; print(tf.__version__)' # for both Python 2 and Python 3
pip list | grep tensorflow will also show the version of Tensorflow installed.
For example, I have installed TensorFlow 0.9.0 in a virtualenv for Python 3. So, I get:
$ python -c 'import tensorflow as tf; print(tf.__version__)'
0.9.0
$ pip list | grep tensorflow
tensorflow (0.9.0)

Almost every normal package in python assigns the variable .__version__ to the current version. So if you want to find the version of some package you can do the following
import a
a.__version__
For tensorflow it will be
import tensorflow as tf
tf.version.VERSION
For old versions of tensorflow (below 0.10), use tf.__version__

If you have installed via pip, just run the following
$ pip show tensorflow
Name: tensorflow
Version: 1.5.0
Summary: TensorFlow helps the tensors flow

import tensorflow as tf
print(tf.VERSION)

For python 3.6.2:
import tensorflow as tf
print(tf.version.VERSION)

If you're using anaconda distribution of Python,
$ conda list | grep tensorflow
tensorflow 1.0.0 py35_0 conda-forge
To check it using Jupyter Notebook (IPython Notebook)
In [1]: import tensorflow as tf
In [2]: tf.__version__
Out[2]: '1.0.0'

For knowing any version of the python library then if your library is installed using the pip then use the following command.
pip show tensorflow
The Output of the above command will be shown below:-
Name: tensorflow
Version: 2.3.0
Summary: TensorFlow is an open source machine learning framework for everyone.
Home-page: https://www.tensorflow.org/
Author: Google Inc.
Author-email: packages#tensorflow.org
License: Apache 2.0
Location: /usr/local/lib/python3.6/dist-packages
Requires: astunparse, wheel, keras-preprocessing, gast, tensorflow-estimator, opt-einsum, tensorboard, protobuf, absl-py, six, wrapt, termcolor, numpy, grpcio, scipy, google-pasta, h5py
Required-by: fancyimpute

I installed the Tensorflow 0.12rc from source, and the following command gives me the version info:
python -c 'import tensorflow as tf; print(tf.__version__)' # for Python 2
python3 -c 'import tensorflow as tf; print(tf.__version__)' # for Python 3
The following figure shows the output:

To get more information about tensorflow and its options you can use below command:
>> import tensorflow as tf
>> help(tf)

On Latest TensorFlow release 1.14.0
tf.VERSION
is deprecated, instead of this use
tf.version.VERSION
ERROR:
WARNING: Logging before flag parsing goes to stderr.
The name tf.VERSION is deprecated. Please use tf.version.VERSION instead.

Easily get KERAS and TENSORFLOW version number -->
Run this command in terminal:
[username#usrnm:~] python3
>>import keras; print(keras.__version__)
Using TensorFlow backend.
2.2.4
>>import tensorflow as tf; print(tf.__version__)
1.12.0

The tensorflow version can be checked either on terminal or console or in any IDE editer as well (like Spyder or Jupyter notebook, etc)
Simple command to check version:
(py36) C:\WINDOWS\system32>python
Python 3.6.8 |Anaconda custom (64-bit)
>>> import tensorflow as tf
>>> tf.__version__
'1.13.1'

Another variation, i guess :P
python3 -c 'print(__import__("tensorflow").__version__)'

python -c 'import tensorflow as tf; print(tf.__version__)' # for Python 2
python3 -c 'import tensorflow as tf; print(tf.__version__)' # for Python 3
Here -c represents program passed in as string (terminates option list)

Tensorflow version in Jupyter Notebook:-
!pip list | grep tensorflow

If you have TensorFlow 2.x:
sess = tf.compat.v1.Session(config=tf.compat.v1.ConfigProto(log_device_placement=True))

For Windows cmd
pip list | FINDSTR tensorflow
OR
pip show tensorflow
For Linux
pip list | grep tensorflow
OR
pip show tensorflow

The version of Tensorflow can simply be checked on jupyter notebook using the following simple steps.
import tensorflow as tf
print(tf.__version__) # for Python 3

Printing python version in human readable format
python -c 'import sys; print(".".join(map(str, sys.version_info[:3])))'