lucidrains/vit-pytorch

By GitHub - 2020-10-05

Description

Implementation of Vision Transformer, a simple way to achieve SOTA in vision classification with only a single transformer encoder, in Pytorch - lucidrains/vit-pytorch

Summary

  • README.md Vision Transformer - Pytorch Implementation of Vision Transformer, a simple way to achieve SOTA in vision classification with only a single transformer encoder, in Pytorch.
  • Significance is further explained in Yannic Kilcher's video.
  • For a Pytorch implementation with pretrained models, please see Ross Wightman's repository here Distillation A recent paper has shown that use of a distillation token for distilling knowledge from convolutional nets to vision transformer can yield small and efficient vision transformers.
  • A pytorch-lightning script is ready for you to use at the repository link above.

 

Topics

  1. Machine_Learning (0.1)
  2. NLP (0.07)
  3. Management (0.01)

Similar Articles

Rethinking Attention with Performers

By Google AI Blog - 2020-10-23

Posted by Krzysztof Choromanski and Lucy Colwell, Research Scientists, Google Research Transformer models have achieved state-of-the-art...

fastai/fastbook

By GitHub - 2020-12-07

The fastai book, published as Jupyter Notebooks. Contribute to fastai/fastbook development by creating an account on GitHub.