Daniel Falbel

I think we usually tune the batch_size to be small as possible while making good estimates of the gradient and that shouldn't depend on the dataset size. Maybe we could...

@sametsoekel Thanks for filing the issue. We currently don't support GPU's in brulee, we'll add support for it soon and let you know here.

In theory it would be possible to use {tfhub} for that. We have implemented steps for text embedding here: https://github.com/rstudio/tfhub/blob/master/R/recipe.R and we could have similar functionality for images. I imagine...

Thanks for reporting! Yes, it should be deleted.

Hi @mikeyEcology We still haven't implemented those for R. To my knowledge there's a `yolo` implementation here: https://github.com/openvolley/ovml/blob/master/R/yolo.R If you want to port them I'd start by implementing the generalized...

Since you already trained the model in python you could jit save it with `torch.jit.save(model, "path/to/file")` and reload it in R with `jit_load()`. This way, the architecture is also serialized,...

Hmm this is tricky. It seems that torchvision register some custom TorchScript operations and they wouldn't be available at the time we `jit_load()` in R. We will need to figure...

Sure I am happy to collaborate! FWIW I was able to load the fasterrcnn model in a not too complicated way by doing the following: - Cloned https://github.com/pytorch/vision - `mkdir...

I think you want to use `jit_load` instead of `load_state_dict` as AFAICT `torch.jit.save` is saving the entire model in the TorchScript format.

Hi, sorry, you currently need to install it from github: ```remotes::install_github("dfalbel/electricity")```