search

sgoldenlab / simba

SimBA (Simple Behavioral Analysis), a pipeline and GUI for developing supervised behavioral classifiers
https://simba-uw-tf-dev.readthedocs.io/
GNU General Public License v3.0
297 stars 143 forks source link

Conflict dependencies #63

Closed flyegg closed 4 years ago

flyegg commented 4 years ago

Describe the bug I pip installed simba-uw-tf in a fresh conda environment but cannot run it. I tried to uninstall and reinstall shapely from conda-forge as suggested by earlier posts but that did not work.

However, I noticed conflict dependencies when I tried to pip check simba-uw-tf.

I didn't test all but here are some example messages I got.

  1. pip check simba-uw-tf tensorflow-gpu 1.14.0 has requirement protobuf>=3.6.1, but you have protobuf 3.6.0. imbalanced-learn 0.7.0 has requirement scikit-learn>=0.23, but you have scikit-learn 0.22.2. deeplabcut 2.0.9 has requirement numpy~=1.14.5, but you have numpy 1.18.1. deeplabcut 2.0.9 has requirement scikit-learn~=0.19.2, but you have scikit-learn 0.22.2

  2. After I install protobuf 3.6.0. using pip install protobuf==3.6.0 tensorflow-gpu 1.14.0 requires protobuf>=3.6.1, but you'll have protobuf 3.6.0 which is incompatible. ...

  3. Then I downgrade tensorflow from 1.14.0 to 1.13.2 simba-uw-tf 1.2.20 has requirement protobuf==3.6.0, but you have protobuf 3.13.0. simba-uw-tf 1.2.20 has requirement tensorflow-gpu==1.14.0, but you have tensorflow-gpu 1.13.2.

I am totally confused .

To Reproduce Steps to reproduce the behavior:

  1. pip install simba-uw-tf
  2. pip check simba-uw-tf
  3. try to adjust the module version as expected

Expected behavior

Screenshots

Desktop (please complete the following information):

Additional context Add any other context about the problem here.

sgoldenlab commented 4 years ago

Hi @flyegg,

The "error" msgs you are seeing after your first points are more "warnings" that shouldn't affect the boot-up and performance of SimBA. However, in my experience, you do need protobuf ==3.6.0, but do not downgrade tensoflow-gpu, stick with 1.14.0.

Can you try the following, if you are running it on Linux (as indicated in your msg), type:

pip uninstall simba-uw-tf, then:

pip install simba-uw-tf-dev

ignore any of the warning msgs you see, then type:

simba

If you see any error msgs, please post me a screenshot.

Thanks

flyegg commented 4 years ago

Thanks, it works like a charm.