sthalles / PyTorch-BYOL

PyTorch implementation of Bootstrap Your Own Latent: A New Approach to Self-Supervised Learning
480 stars 72 forks source link

There should be two MLPS behind the online network and one behind the target network #11

Closed Jason1010 closed 2 years ago

Jason1010 commented 2 years ago

First of all, thank you so much for making this work open source!

There should be two MLP behind the online network in the paper, and there should be one MLP behind the target network in the paper. And in the code, online network has one MLP. But target network has no MLP. Besides, There are also parameter updates between the MLP of the online network and the target network. Is there an oversight here? Or I didn't see it

Jason1010 commented 2 years ago

sorry,I am wrong.