Open Bababalei opened 4 years ago
When I run the Google Colab file, I happened this issue and even though I re-install tensorflow 2.0.0 or manually change from keras import to from tensorflow.keras import, this error still existed
Change the tenslrflow version to the version mentioned in the github. 1.2 something. And it will work.
Change the tenslrflow version to the version mentioned in the github. 1.2 something. And it will work.
so i reseted runtime and then changed tensorflow version with
%tensorflow_version 1.12.0
but then i get another error while running model constructor:
AttributeError Traceback (most recent call last)
so i reseted runtime and then changed tensorflow version with
%tensorflow_version 1.12.0
This does not change tf version to 1.12.0. This magic only change between latest 1.x and 2.x versions. You should manually install 1.12.0 or 1.13.1.
P.S: I use !pip install keras==2.2.4 tensorflow==1.13.1
and notebook works well
I solved this issue by using these command pip uninstall -y tensorflow pip install tensorflow-gpu==1.13.1
still gives this error
from models import FaceTranslationGANInferenceModel
---------------------------------------------------------------------------
ImportError Traceback (most recent call last)
<ipython-input-12-e0f766e65a9c> in <module>()
----> 1 from models import FaceTranslationGANInferenceModel
5 frames
/tensorflow-1.15.2/python3.7/keras/backend/load_backend.py in <module>()
13 from .common import set_image_data_format
14 from .common import normalize_data_format
---> 15 from .common import symbolic, eager
16
17 # Set Keras base dir path given KERAS_HOME env variable, if applicable.
ImportError: cannot import name 'symbolic' from 'keras.backend.common' (/usr/local/lib/python3.7/dist-packages/keras/backend/common.py)
---------------------------------------------------------------------------
NOTE: If your import is failing due to a missing package, you can
manually install dependencies using either !pip or !apt.
To view examples of installing some common dependencies, click the
"Open Examples" button below.
---------------------------------------------------------------------------