guarin
guarin
I believe we should test in vicregloss forward that the features have the same dimension as the grid: https://github.com/lightly-ai/lightly/blob/00461d139d2a4280d91cbbc0d4d32d8766ef4227/lightly/loss/vicregl_loss.py#L97 PR would be very much appreciated!
Oh wow, thanks for the PR! This looks interesting, will have a look ASAP.
Just to make sure, is this already ready for review?
Awesome, let use know once it is ready :)
Btw. we can also run some benchmarks on our side if that helps :)
The failing codecov upload is unrelated to the PR :) Could you run `make format` to apply the autoformatting? See https://github.com/lightly-ai/lightly?tab=readme-ov-file#development for more info.
@jameschapman19 I fixed merged conflicts and some typing issues. I also added a `SSLEYProjectionHead` changed the examples to use the `SSLEYTransform`. I'll now try training the model on imagenet and...
@shaundaley39 can we close this PR?
The model is from this repository https://github.com/google-research/scenic/tree/main/scenic/projects/owl_vit The model code is here: https://github.com/google-research/scenic/blob/30374f97f8fb74a25bdd6220dee2e7f29fb509da/scenic/projects/owl_vit/models.py#L74 And this is a colab to export the model as tensorflow saved model: https://colab.research.google.com/drive/1NI5-5AjxtdbeV9b5uKZPQYwnGxYc5Uak?usp=sharing Hope this helps,...
Hi @fatcat-z, you can find a zip of the model here: https://drive.google.com/file/d/1KOfAgq_D7b-HScZU_H6B18nQtWnHAmlL/view?usp=drive_link