lightly
lightly copied to clipboard
LeJEPA
Hi all,
I have found this today, and I believe it will be a very strong addition to the library!
Looks amazing from my perspective
Hi @MohamedFarag21,
Thank you so much for sharing the paper! We will see if we can put this on our agenda :)
Feel also free to open a PR if you would like to contribute. Loss functions and transforms are usually a great start. Once we have those we can put everything together and run benchmarks. Given that they benchmarked on rather small models and datasets this paper should be simpler to reproduce.