For tensorflow_compression, please install TensorFlow 2.1. （Urgent !)

[DerrickLinus]

I first installed tensorflow-compression using the official command:
python -m pip install tensorflow-compression
But then, when I want to use it, the following error occurred:
import tensorflow_compression as tf
Traceback (most recent call last):
File "/home/jay/workspace/adjscc/tensorflow_compression/init.py", line 20, in
assert _tf_version[0] == 2 and _tf_version[1] == 1
AssertionError

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "", line 1, in
File "/home/jay/workspace/adjscc/tensorflow_compression/init.py", line 22, in
raise RuntimeError(
RuntimeError: For tensorflow_compression, please install TensorFlow 2.1.

The version currently supported by tensorflow-compression has been clearly stated in the github project to be TensorFlow 2.14, but why does it show that TensorFlow 2.1 is required in the initialization file?

[jonaballe]

Hi, it looks like you are importing an old version TFC from your workspace rather than the one you installed with pip. When you "import tensorflow_compression", Python will first search in your current directory.

[DerrickLinus]

Thank you for your response. Could you please clarify the proper way to import tensorflow-compression? I've created a conda virtual environment named 'tf-gpu' where I installed the package, but I'm still encountering the import error mentioned earlier. Does this indicate an environment configuration issue or potential path misalignment?

I'm currently blocked from proceeding with my experiments due to this problem and would greatly appreciate your guidance on resolving this dependency conflict. Please let me know if you need additional system details or environment specifications to better diagnose the issue.

[jonaballe]

Do rm -r /home/jay/workspace/adjscc/tensorflow_compression

[DerrickLinus]

Thank you so much for your guidance! I'm pleased to confirm the issue has been fully resolved.