-
Notifications
You must be signed in to change notification settings - Fork 114
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Update requirements #127
Update requirements #127
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thank you @jdeschamps for cleaning up!
So far I only glanced over it and it looks all reasonable to me. I will check it out and run the tests later today.
We also need to update the installation instructions. I will test this later today! This page recommends to also specify |
It is probably a good idea to follow this recommendation. Could you ping me once you updated the instructions? Then I can test-run it from start to finish ✌️ |
Done (0c0f790). I chose to follow the TF guidelines and link to it. We could also just say use |
Thank you @jdeschamps for all the work! I was able to install everything and run the tests and example notebooks as they are. I did not run the functional test. They just take too long. I also tested the N2V2 functionality by setting
Training and prediction works well for the 2D case (in 3D the model is not initialized with an |
@tibuch oh good catch. I have also not tried the functional tests, I have excluded them from tox testenv. I will start them on a virtual machine today and check them tomorrow. One more thing. It seems that on my RedHat machine, I need to set the |
This should be fixed by running this once in the acitvate env:
|
I actually had missed your comments because I didn't refresh the page. Perfect! |
The functional tests run through without errors btw. |
FIxing exporting model weight
I merged here the fix for the model export (#128). However there is still the bug that we cannot save twice in a row. I believe this is due to the way CSBDeep saves the model to Since the export works, I would suggest to create an issue about this bug to keep track of it and see what becomes of it in the next CSBDeep version. In the mean time since it does not prevent users from using N2V notebooks, I would go ahead with the new version release. |
I tested all notebooks with TF 2.7 and 2.10 (latest), and CSBDeep 0.7.2 (latest) in Python 3.9.
Here is roughly how I installed the environment: