Skip to content
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.

Sign up for GitHub

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

Jump to bottom

[GSoC] TF 2.0 Custom ResNet Training #79

Closed
wants to merge 8 commits into from
Closed

[GSoC] TF 2.0 Custom ResNet Training #79

wants to merge 8 commits into from

Conversation

Vishal-V
Copy link
Contributor

@Vishal-V Vishal-V commented Jul 22, 2019 •
edited
Loading

  • Custom ResNet model for the TinyImageNet dataset without transfer learning.
  • Efficient model with tf.keras.layers.SeparableConv2D and Resizing strategy with ECLR to train with colab-like resource constrained environments.
  • TF 2.0 compatible model and associated training and regularization strategies to be colab friendly.
  • Uses tf.keras and eager by default with option to train with tf.GradientTape() eagerly.
  • Graph mode training with tf.keras.ImageDataGenerator and custom image pre-processing functions.

@googlebot googlebot added the cla: yes CLA has been signed label Jul 22, 2019
@review-notebook-app
Copy link

Check out this pull request on ReviewNB: https://app.reviewnb.com/tensorflow/examples/pull/79

You'll be able to see visual diffs and write comments on notebook cells. Powered by ReviewNB.

@MarkDaoust MarkDaoust self-requested a review July 29, 2019 23:33
@MarkDaoust MarkDaoust self-assigned this Jul 29, 2019
MarkDaoust
MarkDaoust previously approved these changes Jul 29, 2019
View reviewed changes
Copy link
Member

@MarkDaoust MarkDaoust left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Overall this is in good shape.

I don't see any 'Superconvergence'.
Here are a couple of little things you could fix.

But aside from the bugs in the notebook I think this is good to go.

community/en/tiny_imagenet_custom_resnet.ipynb Outdated Show resolved Hide resolved
community/en/tiny_imagenet_custom_resnet.ipynb Outdated Show resolved Hide resolved
community/en/tiny_imagenet_custom_resnet.ipynb Outdated Show resolved Hide resolved
community/en/tiny_imagenet_custom_resnet.ipynb Outdated Show resolved Hide resolved
community/en/tiny_imagenet_custom_resnet.ipynb Outdated Show resolved Hide resolved
community/en/tiny_imagenet_custom_resnet.ipynb Outdated Show resolved Hide resolved
community/en/tiny_imagenet_custom_resnet.ipynb Outdated Show resolved Hide resolved
tensorflow_examples/models/custom_resnet/load_data.py Outdated Show resolved Hide resolved
@MarkDaoust
Copy link
Member

Cool, it sounds like you've got this under control, I think we're good to merge this once you have those changes uploaded.

@Vishal-V
Copy link
Contributor Author

Vishal-V commented Aug 22, 2019 •
edited
Loading

@MarkDaoust I did not mean to dismiss the review. The commit I added automatically raised that message. The notebook is ready for review.

@Vishal-V Vishal-V marked this pull request as ready for review August 22, 2019 15:55
@lamberta
Copy link
Member

As mentioned in the other PR, this community/ directory has been deprecated. We're no longer accepting pull requests here. Unclear where the rest of this PR stands, though.

@MarkDaoust
Copy link
Member

Sorry this didn't work out.

@MarkDaoust MarkDaoust closed this Jan 12, 2023
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@MarkDaoust MarkDaoust MarkDaoust left review comments

@yashk2810 yashk2810 yashk2810 left review comments

@adammichaelwood adammichaelwood Awaiting requested review from adammichaelwood

@lamberta lamberta Awaiting requested review from lamberta

@wolffg wolffg Awaiting requested review from wolffg

Assignees

@MarkDaoust MarkDaoust

Labels
cla: yes CLA has been signed
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
5 participants
@Vishal-V @MarkDaoust @lamberta @yashk2810 @googlebot