pytorch / pytorch/vision

pre-activation ResNet

Open
#205 5 comments 13 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
17.9k
Forks
7.3k
Avg merge
1d 15h
Merged PRs (30d)
13

Description

Is there any interest in implementing the "pre-activation" version of ResNet, as described in

https://arxiv.org/pdf/1603.05027.pdf

and which has been show to have better performance on CIFAR 10/100 and Imagenet? It's a relatively minor modification of the ResNet version that's already implemented -- just changing the order of some of the operations in the residual block.

I'm happy to implement and submit a PR, though I don't really have the resources to train an Imagenet model ATM.

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start from the existing ResNet implementation and compare its residual block with the pre-activation design described in the linked paper. Apply the operation-order change for the existing CIFAR-10/100 and ImageNet model variants, then verify the model behavior with the repository's available checks; ImageNet training is not required by the issue.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
computer-vision, machine-learning
Issue type
Feature
Difficulty
3/5
Estimated time
1-2 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
38/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.