Skip to content
This repository has been archived by the owner on Mar 21, 2024. It is now read-only.

Upgrade to PyTorch 1.7 and Lightning 1.1.6 #370

Closed
ant0nsc opened this issue Jan 28, 2021 · 2 comments · Fixed by #411
Closed

Upgrade to PyTorch 1.7 and Lightning 1.1.6 #370

ant0nsc opened this issue Jan 28, 2021 · 2 comments · Fixed by #411
Labels
architecture Anything around possible extensions or re-structuring of code

Comments

@ant0nsc
Copy link
Contributor

ant0nsc commented Jan 28, 2021 •
edited by azure-boards bot
Loading

  • Pytorch 1.7 will allow to remove np.count_nonzero in get_number_of_voxels_per_class
  • Lightning 1.1.6 will enable correctly logging the SecondsPerBatch metric on the training set (at the moment it is mysteriously swallowed, even though logged)

AB#3860
@ant0nsc
Copy link
Contributor Author

ant0nsc commented Jan 28, 2021

Add more checks in _test_model_train: This should check that all metrics are both in train and test metrics, and both files have as many rows as there are epochs.

@ant0nsc ant0nsc added the architecture Anything around possible extensions or re-structuring of code label Feb 2, 2021
@ant0nsc ant0nsc linked a pull request Mar 29, 2021 that will close this issue
Upgrade to Pytorch 1.8 #411
Merged
@ant0nsc
Copy link
Contributor Author

ant0nsc commented Mar 29, 2021

Most work done in #411. Change to torch.count_nonzero in #417

@ant0nsc ant0nsc closed this as completed Mar 29, 2021
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Assignees
No one assigned
Labels
architecture Anything around possible extensions or re-structuring of code
Projects
None yet
Milestone
No milestone
Development

Successfully merging a pull request may close this issue.

Upgrade to Pytorch 1.8 microsoft/InnerEye-DeepLearning
1 participant
@ant0nsc