mlr3torch

mlr3torch copied to clipboard

TODOs

1

* How to use predefined architectures * How slow is it?

sebffischer

the nn_linear_0 in the readme should not be necessary and is not transparent

sebffischer

sebffischer

Implement: https://torch.mlverse.org/docs/reference/torch_adaptive_avg_pool1d.html?q=adaptive#null for 1d, 2d and 3d. This is similar to: https://github.com/mlr-org/mlr3torch/blob/main/R/PipeOpTorchAvgPool.R

sebffischer

sebffischer

https://github.com/mlr-org/mlr-org.r-universe.dev/blob/main/packages.json

sebffischer

sebffischer

sebffischer

* GitHub has GPU support on [their roadmap](https://github.com/github/roadmap/issues/505) but it is not clear when this will be available.

sebffischer

sebffischer