lucidrains / byol-pytorch

Usable Implementation of "Bootstrap Your Own Latent" self-supervised learning, from Deepmind, in Pytorch
MIT License
1.76k stars 248 forks source link

Example Code Parameters #35

Closed JosephDiPalma closed 4 years ago

JosephDiPalma commented 4 years ago

I believe the example code is wrong. The optimizer should only take the parameters from the online encoder and predictor and not all of the parameters like it currently does.

lucidrains commented 4 years ago

@JosephDiPalma Oh yes, I do believe you are right! The latest change should address this issue https://github.com/lucidrains/byol-pytorch/releases/tag/0.3.2 :pray: