recommenders VS pytorch-lightning

recommenders – Best Practices on Recommendation Systems

I would like to do a benchmarking on the Microsoft models SVD, SAR and NMF (available here: https://github.com/microsoft/recommenders) but with this input data I get a precision and recall close to zero. Any ideas how I can improve this? For SVD and NMF (surprise library) the model wants a rating input that is normally distributed, which it not the case for my binary data where the transactions all have a rating of 1.

Then I tried to find some more advanced models and I found this really good list and in there I found the Microsoft one. So it's' where we are now, which a bunch of different models and not a documentation/tutorials out there.

https://see.stanford.edu/Course/CS229 https://lightning.ai/ https://www.youtube.com/watch?v=00s9ireCnCw&t=57s https://towardsdatascience.com/

I've been wondering on what is the recommended method of saving logits/activations using PyTorch Lightning. I've looked at Callbacks, Loggers and ModelHooks but none of the use-cases seem to be for this kind of activity (even if I were to create my own custom variants of each utility). The ModelCheckpoint Callback in its utility makes me feel like custom Callbacks would be the way to go but I'm not quite sure. This closed GitHub issue does address my issue to some extent.

Our codebase for the diffusion models builds heavily on OpenAI's ADM codebase , lucidrains, Stable Diffusion, Lightning and Hugging Face. Thanks for open-sourcing!

For lightweight use, pytorch-lightning is too heavy, and its source code will be very difficult for beginners to read, at least for me.

For the A100 GPU, theoretical performance is the same for FP16/BF16 and both rely on the same number of bits, meaning memory should be the same. However since it's quite newly added to PyTorch, performance seems to still be dependent on underlying operators used (pytorch lightning debugging in progress here).