I've been trying to use tensorflow for two days now installing and reinstalling it over and over again in python2.7 and 3.4. No matter what I do, I get this error message when trying to use tensorflow.placeholder()
It's very boilerplate code:
tf_in = tf.placeholder("float", [None, A]) # Features No matter what I do I always get the trace back:
Traceback (most recent call last): File "/home/willim/PycharmProjects/tensorflow/tensorflow.py", line 2, in <module> import tensorflow as tf File "/home/willim/PycharmProjects/tensorflow/tensorflow.py", line 53, in <module> tf_in = tf.placeholder("float", [None, A]) # Features AttributeError: 'module' object has no attribute 'placeholder' Anyone know how I can fix this?
121 Answers
If you have this error after an upgrade to TensorFlow 2.0, you can still use 1.X API by replacing:
import tensorflow as tf by
import tensorflow.compat.v1 as tf tf.disable_v2_behavior() Solution: Do not use "tensorflow" as your filename.
Notice that you use tensorflow.py as your filename. And I guess you write code like:
import tensorflow as tf Then you are actually importing the script file "tensorflow.py" that is under your current working directory, rather than the "real" tensorflow module from Google.
Here is the order in which a module will be searched when importing:
6
The directory containing the input script (or the current directory when no file is specified).
PYTHONPATH (a list of directory names, with the same syntax as the shell variable PATH).
The installation-dependent default.
It happened to me too. I had tensorflow and it was working pretty well, but when I install tensorflow-gpu along side the previous tensorflow this error arose then I did these 3 steps and it started working with no problem:
- I removed tensorflow-gpu, tensorflow, tensorflow-base packages from Anaconda. Using.
conda remove tensorflow-gpu tensorflow tensorflow-base
- re-installed tensorflow. Using
conda install tensorflow
Instead of tf.placeholder(shape=[None, 2], dtype=tf.float32) use something like tf.compat.v1.placeholder(shape=[None, 2], dtype=tf.float32) if you don't want to disable v2 completely.
import tensorflow.compat.v1 as tf tf.disable_v2_behavior() works. I am using Python 3.7 and tensorflow 2.0.
It appears that .placeholder() , .reset_default_graph() , and others were removed with version 2. I ran into this issue using Docker image: tensorflow/tensorflow:latest-gpu-py3 which automatically pulls the latest version. I was working in 1.13.1 and was 'upgraded to 2' automatically and started getting the error messages. I fixed this by being more specific with my image: tensorflow/tensorflow:1.13.1-gpu-py3.
Avoid using the below striked out statement in tensorflow=2.0
i̶m̶p̶o̶r̶t̶ ̶t̶e̶n̶s̶o̶r̶f̶l̶o̶w̶ ̶a̶s̶ ̶t̶f̶ ̶x̶ ̶=̶ ̶t̶f̶.̶p̶l̶a̶c̶e̶h̶o̶l̶d̶e̶r̶(̶s̶h̶a̶p̶e̶=̶[̶N̶o̶n̶e̶,̶ ̶2̶]̶,̶ ̶d̶t̶y̶p̶e̶=̶t̶f̶.̶f̶l̶o̶a̶t̶3̶2̶)̶
You can disable the v2 behavior by using the following code
This one is perfectly working for me.
import tensorflow.compat.v1 as tf tf.disable_v2_behavior() x = tf.placeholder(shape=[None, 2], dtype=tf.float32) I also got the same error. May be because of the version of tensorflow. After installing tensorflow 1.4.0, I got relief from the error.
pip install tensorflow==1.4.0 If you are using TensorFlow 2.0, then some code developed for tf 1.x may not code work. Either you can follow the link :
or you can install a previous version of tf by pip3 install tensorflow==version
Import the old version of tensorflow instead of the new version
import tensorflow.compat.v1 as tf tf.disable_v2_behavior()
You need to use the keras model with tensorflow 2, as here
import tensorflow as tf from tensorflow.python.keras.layers import Input, Embedding, Dot, Reshape, Dense from tensorflow.python.keras.models import Model Recent version 2.0 does not support placeholder. I uninstalled 2.0 using command: conda remove tensorflow. then I installed 1.15.0 using command: conda install -c conda-forge tensorflow=1.15.0. 1.15 is latest in version 1 series. You can change as per you wish and requirement. For seeing all version, use command: conda search tensorflow. It worked for Anaconda3 in Windows.
Try this:
pip install tensorflow==1.14 or this (if you have GPU):
pip install tensorflow-gpu==1.14 If you get this on tensorflow 2.0.0+, it's very likely because the code isn't compatible with the newer version of tensorflow.
To fix this, run the tf_upgrade_v2 script.
tf_upgrade_v2 --infile=YOUR_SCRIPT.py --outfile=YOUR_SCRIPT.py Faced same issue on Ubuntu 16LTS when tensor flow was installed over existing python installation.
Workaround: 1.)Uninstall tensorflow from pip and pip3
sudo pip uninstall tensorflow sudo pip3 uninstall tensorflow 2.)Uninstall python & python3
sudo apt-get remove python-dev python3-dev python-pip python3-pip 3.)Install only a single version of python(I used python 3)
sudo apt-get install python3-dev python3-pip 4.)Install tensorflow to python3
sudo pip3 install --upgrade pip for non GPU tensorflow, run this command
sudo pip3 install --upgrade tensorflow for GPU tensorflow, run below command
sudo pip3 install --upgrade tensorflow-gpu Suggest not to install GPU and vanilla version of tensorflow
The error shows up because we are using tensorflow version 2 and the command is from version 1. So if we use:
tf.compat.v1.summary.(method_name) It'll work
Please take a look at the Migrate your TensorFlow 1 code to TensorFlow 2.
These codes:
import tensorflow as tf tf_in = tf.placeholder("float", [None, A]) # Features need to be migrated in TensorFlow 2 as below:
import tensorflow as tf import tensorflow.compat.v1 as v1 tf_in = vi.placeholder("float", [None, A]) # Features Because you cant use placeholder in tensflow2.0version, so you need to use tensflow1*, or you need to change your code to fix tensflow2.0
I had the same problem before after tried to upgrade tensorflow, I solved it by reinstalling Tensorflow and Keras.
pip uninstall tensorflow
pip uninstall keras
Then:
pip install tensorflow
pip install keras
The problem is with TensorFlow version; the one you are running is 2.0 or something above 1.5, while placeholder can only work with 1.4.
So simply uninstall TensorFlow, then install it again with version 1.4 and everything will work.
It may be the typo if you incorrectly wrote the placeholder word. In my case I misspelled it as placehoder and got the error like this: AttributeError: 'module' object has no attribute 'placehoder'
