mlr3torch
mlr3torch copied to clipboard
mlr-org
Deep learning framework for the mlr3 ecosystem based on torch
Don't fully train a learner, just show the generated network (`model_descriptor_to_module`).
This PR adds an experimental feature that allows to convert a `torch::dataset` to an `mlr3::Task`. Essentially, the `torch::dataset` is converted to a `data.table` consisting only of `lazy_tensor` columns (including the...
this causes opening of pdf device.
For some neural networks, `jit_trace`ing works, for others it does not. Currently, each learner has the `jit_trace` parameter. We should: * [ ] add a simple test whether jit tracjing...
Should cover the most important parameters to speed up training and what kind of performance there is to be expected.
Call it `po("trafo_fn")` and `po("augment_fn")`.
https://arxiv.org/pdf/2106.11959 Steps: 1. Allow transformation in internal tune space, otherwise we can't have one tuning space (that includes early stopping). Also need to change part of the mlr3 book 2....