https://github.com/oke-aditya/pytorch_paper_implementations

Deep Learning Paper Implementations in PyTorch
https://github.com/oke-aditya/pytorch_paper_implementations

deep-learning deep-learning-papers implementation-of-research-paper pytorch

Last synced: 16 days ago
JSON representation

Deep Learning Paper Implementations in PyTorch

• Host: GitHub
• URL: https://github.com/oke-aditya/pytorch_paper_implementations
• Owner: oke-aditya
• License: gpl-3.0
• Created: 2020-06-18T11:49:04.000Z (over 4 years ago)
• Default Branch: master
• Last Pushed: 2020-11-11T17:39:39.000Z (about 4 years ago)
• Last Synced: 2024-10-12T11:06:50.921Z (about 1 month ago)
• Topics: deep-learning, deep-learning-papers, implementation-of-research-paper, pytorch
• Language: Python
• Homepage:
• Size: 43.9 KB
• Stars: 17
• Watchers: 2
• Forks: 2
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Deep Learning Paper Implementations

Implements deep learning papers. It contains basic papers such as activations, CNNs to advanced papers such as encoder-decoder etc.

It also contains links to paper from which it is taken.

My Aim is to learn to write research code and easily reproduce it from papers.

Some implementations are mine, most will be taken from various GitHub repos.

My aim here is to learn, reproduce work and understand nuances.

# Why this Repo ?

- It presents easily reusable code for Deep Learning Papers.

- It contains ```*.py``` and easy to reproduce code.

- It becomes harder for people to open ```*.ipynb``` files and decode stuff and reuse from them. I have created functions and classes which is best practice of writing code.

- Code is written in Pytorch so that most community can benefit.

- Contributions for tensorflow code are welcome.

# What this repo is not for ?

- It is not a repo to showcase my DL skills or show off any implementations.

- Not to beat and achieve benchmarks.

- Not a repo to prove points how one paper is better than other. It depends on context where you apply DL. You won't fit a neural network when decision tree might work.

- Not a repo to showcase how papers keep battling for SOTA. I want to reimplement stuff not create a benchmarking repo.

# FAQs

## So most code here is reimplemenatation nothing new ?

- Yes as of now. I'm not a great DL researcher / scientist; just a student.

- I would suggest use https://github.com/rwightman/pytorch-image-models this repo for models.

- I have tried to learn from its implmenatations as well.

## Can I copy paste some of your code from here ?

- Yes you can. It is just reimplementation. But I do not guarentee if that will help you to get SOTA.

## I found a bug what to do ?

- Raise a PR

## I want to contribute with some implementation?

- Document and comment the code well. Raise a PR.

## Can I use Fastai, or Keras to contribute?

- No, please do not for this repo is not for such frameworks that have too much built in power.

- But yes please use tensorflow v2 or Pytorch or Pytorch-lightning.