Search results

Improve docu on output heads and losses.

Status: Open.

#384 In mlr-org/mlr3torch;

· sebffischer opened on Apr 16, 2025

<code>nn_graph</code> should not require specifying input shapes

Status: Open.

#383 In mlr-org/mlr3torch;

· sebffischer opened on Apr 14, 2025

Improve documentation for <code>PipeOpTorchIngress</code>

Status: Open.

#380 In mlr-org/mlr3torch;

· sebffischer opened on Apr 10, 2025

Better options for debugging a deep learning graph

Status: Open.

#379 In mlr-org/mlr3torch;

· sebffischer opened on Apr 10, 2025

Maybe consider specialized dataloader when <code>tensor_dataset = TRUE</code>

Status: Open.

#378 In mlr-org/mlr3torch;

· sebffischer opened on Apr 7, 2025

LearnerTorchModel should be added to dictionary of learners

Status: Open.

#376 In mlr-org/mlr3torch;

· sebffischer opened on Apr 3, 2025

tensor size errors for bce_with_logits loss / net which outputs 1d tensor

Status: Open.

#373 In mlr-org/mlr3torch;

· tdhock opened on Apr 2, 2025

Add <code>po("trafo_custom")</code> that wraps or replaces <code>pipeop_preproc_torch</code>

Status: Open.

#368 In mlr-org/mlr3torch;

· sebffischer opened on Mar 21, 2025

<code>PipeOpTorchModel</code> should expose fields <code>callbacks</code>, <code>optimizer</code> and <code>loss</code> so they can be changed afterwards

Status: Open.

#366 In mlr-org/mlr3torch;

· sebffischer opened on Mar 21, 2025

example of creating your own task?

Status: Open.

#361 In mlr-org/mlr3torch;

· tdhock opened on Feb 17, 2025

Add feature that allows to train a <code>LearnerTorch</code> on task where the target is also a lazy tensor

Status: Open.

#357 In mlr-org/mlr3torch;

· sebffischer opened on Feb 14, 2025

Better error messages when network diverges and yields NA predictions

Status: Open.

#353 In mlr-org/mlr3torch;

· sebffischer opened on Feb 8, 2025