search

sezan92 / computer-vision-course

This repo is the homebase of a community driven course on Computer Vision with Neural Networks. Feel free to join us on the Hugging Face discord: hf.co/join/discord
MIT License
1 stars 0 forks source link

Create convnext.mdx #5

Closed JvThunder closed 8 months ago

sezan92 commented 8 months ago

@JvThunder I am unable to review for now. I have assigned @sitamgithub-MSIT and @albertkao227

albertkao227 commented 8 months ago

Hi @JvThunder
I think this is good, we can go from here and add more topics.

If we want to make a stronger point in the intro, I have a small suggestion -- while transformer architecture has helped improve performance of multiple CV tasks, there was no theory or clear data to support ViT being superior to CNN (complexity, flexibility, etc). Based on first principles, CNN should be able to achieve what ViT could, if not better.

Just an idea, not necessarily important. Thanks for sharing.