ModuleNotFoundError: No module named 'tensorflow'

[songzhibo1]

Hello, thank you very much for sharing the code. I encountered some errors while using Docker to pull the code and run bash build_obf.sh. I would like to ask for your guidance on how to resolve this issue. Your help would be greatly appreciated!

Thank you again for your support!

ERROR: An error occurred during the fetch of repository 'tf_runtime':
Traceback (most recent call last):
File "/Code275/tensorflow-2.9.1/third_party/repo.bzl", line 64, column 33, in _tf_http_archive_impl
ctx.download_and_extract(
Error in download_and_extract: java.io.IOException: Error downloading [https://storage.googleapis.com/mirror.tensorflow.org/github.com/tensorflow/runtime/archive/093ed77f7d50f75b376f40a71ea86e08cedb8b80.tar.gz, https://github.com/tensorflow/runtime/archive/093ed77f7d50f75b376f40a71ea86e08cedb8b80.tar.gz] to /root/.cache/bazel/_bazel_root/a5a5071651a6c349d639f25e016288e1/external/tf_runtime/temp5368221067054860666/093ed77f7d50f75b376f40a71ea86e08cedb8b80.tar.gz: connect timed out
ERROR: /Code275/tensorflow-2.9.1/WORKSPACE:11:14: fetching _tf_http_archive rule //external:tf_runtime: Traceback (most recent call last):
File "/Code275/tensorflow-2.9.1/third_party/repo.bzl", line 64, column 33, in _tf_http_archive_impl
ctx.download_and_extract(
Error in download_and_extract: java.io.IOException: Error downloading [https://storage.googleapis.com/mirror.tensorflow.org/github.com/tensorflow/runtime/archive/093ed77f7d50f75b376f40a71ea86e08cedb8b80.tar.gz, https://github.com/tensorflow/runtime/archive/093ed77f7d50f75b376f40a71ea86e08cedb8b80.tar.gz] to /root/.cache/bazel/_bazel_root/a5a5071651a6c349d639f25e016288e1/external/tf_runtime/temp5368221067054860666/093ed77f7d50f75b376f40a71ea86e08cedb8b80.tar.gz: connect timed out
ERROR: no such package '@tf_runtime//': java.io.IOException: Error downloading [https://storage.googleapis.com/mirror.tensorflow.org/github.com/tensorflow/runtime/archive/093ed77f7d50f75b376f40a71ea86e08cedb8b80.tar.gz, https://github.com/tensorflow/runtime/archive/093ed77f7d50f75b376f40a71ea86e08cedb8b80.tar.gz] to /root/.cache/bazel/_bazel_root/a5a5071651a6c349d639f25e016288e1/external/tf_runtime/temp5368221067054860666/093ed77f7d50f75b376f40a71ea86e08cedb8b80.tar.gz: connect timed out
INFO: Elapsed time: 64.286s
INFO: 0 processes.
FAILED: Build did NOT complete successfully (0 packages loaded)
Fetching @llvm-raw; fetching 56s
Fetching ...vm-raw; Extracting /root/.cache/bazel/_bazel_root/a5a5071651a6c349d639f25e
016288e1/external/llvm-raw/temp17482780817171387685/1cb299165c859533e22f2ed05eb2abd5071544
df.tar.gz 6s
build.sh: line 3: ./bazel-bin/tensorflow/tools/pip_package/build_pip_package: No such file or directory
[1000, 224, 224, 3]
Data type of this model is float32
Estimated count of arithmetic ops: 0 ops, equivalently 0 MACs
Found existing installation: tensorflow 2.9.1
Uninstalling tensorflow-2.9.1:
Successfully uninstalled tensorflow-2.9.1
WARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with the system package manager. It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv
WARNING: Requirement '/tmp/tensorflow_pkg/tensorflow-2.9.1-cp38-cp38-linux_x86_64.whl' looks like a filename, but the file does not exist
Processing /tmp/tensorflow_pkg/tensorflow-2.9.1-cp38-cp38-linux_x86_64.whl
ERROR: Could not install packages due to an OSError: [Errno 2] No such file or directory: '/tmp/tensorflow_pkg/tensorflow-2.9.1-cp38-cp38-linux_x86_64.whl'

Traceback (most recent call last):
File "test_model.py", line 5, in
import tensorflow as tf
ModuleNotFoundError: No module named 'tensorflow'

[zhoumingyi]

It looks like the error is in the customized TFLite compilation. You can follow the steps to find the error message:

You can run the commands one by one (these commands can be found in the building scripts):

generate the obfuscation model and modified source code of TFLite

python obfuscation.py --model_name=fruit --extra_layer=30 --shortcut=30

build the customized tflite from the modified source code, the problem is usually caused by this step. Note that if your machine has enough RAM (>32 GB), you can remove the option "--jobs=14", it defines the maximal nb. of compiling jobs:

cd ./tensorflow-2.9.1/
bazel build --jobs=14 //tensorflow/tools/pip_package:build_pip_package

generate python package

./bazel-bin/tensorflow/tools/pip_package/build_pip_package /tmp/tensorflow_pkg
cd ..

uninstall the original TFLite

python -m pip uninstall -y tensorflow

install the customized TFLite

python -m pip install /tmp/tensorflow_pkg/tensorflow-2.9.1-cp38-cp38-linux_x86_64.whl

test

python test_model.py --model_name=fruit

When you identify the step causing the problem, you can post the error message here. And I can try my best to help you.

[carlok]

I confirm that the instructions on the README.md are not enough: when you reach the

bash build_obf.sh

point, you get an error: it disintalls tensorflow but it does not install the local whl

so I tried to follow this previous 5-comment

bazel build --jobs=14 //tensorflow/tools/pip_package:build_pip_package

but it was not able to complete the build due to a couple of 404:

WARNING: Download from https://storage.googleapis.com/mirror.tensorflow.org/github.com/tensorflow/runtime/archive/093ed77f7d50f75b376f40a71ea86e08cedb8b80.tar.gz failed: class java.io.FileNotFoundException GET returned 404 Not Found
WARNING: Download from https://mirror.bazel.build/github.com/bazelbuild/rules_cc/archive/081771d4a0e9d7d3aa0eed2ef389fa4700dfb23e.tar.gz failed: class java.io.FileNotFoundException GET returned 404 Not Found

so AFAIK it does not seems to work due to old instructions.

[zhoumingyi]

I confirm that the instructions on the README.md are not enough: when you reach the

bash build_obf.sh

point, you get an error: it disintalls tensorflow but it does not install the local whl

so I tried to follow this previous 5-comment

bazel build --jobs=14 //tensorflow/tools/pip_package:build_pip_package

but it was not able to complete the build due to a couple of 404:

WARNING: Download from https://storage.googleapis.com/mirror.tensorflow.org/github.com/tensorflow/runtime/archive/093ed77f7d50f75b376f40a71ea86e08cedb8b80.tar.gz failed: class java.io.FileNotFoundException GET returned 404 Not Found
WARNING: Download from https://mirror.bazel.build/github.com/bazelbuild/rules_cc/archive/081771d4a0e9d7d3aa0eed2ef389fa4700dfb23e.tar.gz failed: class java.io.FileNotFoundException GET returned 404 Not Found

so AFAIK it does not seems to work due to old instructions.

The environment is indeed a problem for reproducing this work because the source codes of TensorFlow and its dependencies evolve very quickly. Do you try to use Docker? I will check it and see what the problem is.

[zhoumingyi]

I confirm that the instructions on the README.md are not enough: when you reach the

bash build_obf.sh

point, you get an error: it disintalls tensorflow but it does not install the local whl

so I tried to follow this previous 5-comment

bazel build --jobs=14 //tensorflow/tools/pip_package:build_pip_package

but it was not able to complete the build due to a couple of 404:

WARNING: Download from https://storage.googleapis.com/mirror.tensorflow.org/github.com/tensorflow/runtime/archive/093ed77f7d50f75b376f40a71ea86e08cedb8b80.tar.gz failed: class java.io.FileNotFoundException GET returned 404 Not Found
WARNING: Download from https://mirror.bazel.build/github.com/bazelbuild/rules_cc/archive/081771d4a0e9d7d3aa0eed2ef389fa4700dfb23e.tar.gz failed: class java.io.FileNotFoundException GET returned 404 Not Found

so AFAIK it does not seems to work due to old instructions.

It seems that some dependencies for compiling the old TF's source code have been deprecated now. If it is, we will try to solve it by updating the code to support the recent TF or just packing some essential dependencies into the Docker image.

[carlok]

Hello.
Yes, I tried using Docker, the way that README suggested.
Thank you.
I'd suggest not just to fix the "deprecated" parts but to update the Dockerfile itself to avoid having to follow extra steps after getting the image, as it happens now.