Lime-Cakes
Lime-Cakes
I've had some success running this on google colab, by just git cloning it and running with gpu. I can share the notebook, but it's better to wait for the...
> > I've had some success running this on google colab, by just git cloning it and running with gpu. I can share the notebook, but it's better to wait...
Is there any update with this?
I found an alternative, easier but hacky method. Still testing it at the moment. But it should work allow saving flax unet/vae/any model that uses `convert_pytorch_state_dict_to_flax` in pytorch format. Instead...
> thanks @Lime-Cakes heart this is super cool news fire if you need tester I can test > Hi @Lime-Cakes, that's a perfect approach in my opinion :) [Pull request...
I just tested the issue today and realize the error changed. Instead of taking around 470 seconds, it now takes 90 seconds. It now ends after one set of 6...
Yes, it seems fixed in newer version. Been ages and I forgot about this issue.
I'm interested in the jax version as well! Ideally, the same training code used, so it'd be possible to reproduce the training process.
> Hey, I've implemented a Jax version that I can release soon when I'm finished with my project :)! Looking forward to it! Hope it's released soon
If there is concern about model release, could paper/technical report/code be released first? Information like training method, hyper parameter, and model architecture is of interest to many people. Releasing information...