ModuleNotFoundError: No module named 'tensorflow_gnn.proto.graph_schema_pb2' about gnn HOT 12 CLOSED

To get around this error, clone the main TensorFlow repository, then manually compile the protocol buffer files in the proto directory. Assuming you are working from your home directory:

pip install grpcio grpcio-tools
git clone https://github.com/tensorflow/tensorflow.git
cd tensorflow_gnn/tensorflow_gnn/proto
python -m grpc_tools.protoc -I . -I ~/tensorflow --python_out=. graph_schema.proto
python -m grpc_tools.protoc -I . -I ~/tensorflow --python_out=. examples.proto

from gnn.

Same error when run this example https://github.com/yarri-oss/gnn/blob/main/examples/citeseer/graph_tensor_example.py

from gnn.

@joshcarty Do you have this issue when you run gat_v2_test.py? Thank you.

from gnn.

Answered here.

I'd also be wary of using the above fork - it's from an older version of the package and quite a few changes have been made since, including the removal of those examples.

from gnn.

Had run the command before.

should have no error in the highlight box in pycharm if everything installed correctly?

from gnn.

The error suggests that there was an issue with the installation, as the modules that need to be built from Protobufs are not available.

Do you want to pip install the package again and show us the output of the install if it looks like there was a problem?

from gnn.

Thank you. The reason is my tf can't install successfully.

1. My cpu does not support avx, so I comment #'tensorflow>=2.7.0', in setup.py and build tf by myself.
2. Use pip to install my tf but got ERROR: tensorflow-2.7.0-cp39-cp39-linux_x86_64.whl is not a supported wheel on this platform.
3. Use sudo pip to install my tf successfully.
4. Why tf build in my OS can't install to my Os by general user but can install it successfully by super user?

from gnn.

It seems general user use /home/sidney/.local/bin/pip3.8 and super user use /usr/local/bin/pip3.9 cause the problem.

from gnn.

ssh://[email protected]:22/usr/bin/python -u /home/sidney/py_proj/tf-gnn/tensorflow_gnn/graph/keras/layers/gat_v2_test.py
free(): invalid pointer

Process finished with exit code 134

this error occur after all installed correctly, what's the problem?

from gnn.

So the tf is not installed correctly? How to solve this problem then?

from gnn.

Run gat_v2_test.py successfully now. But need copy the generated file graph_schema_pb2.py to the proto folder, is it normal?

from gnn.

Thanks. This just generate graph_schema_pb2.py from graph_schema.proto in proto folder, right? it is generated in another place when install tensorflow_gnn, and just need to copy it to proto folder.

from gnn.