Ethan Harris

Request to un-pin the PyTorch Lightning Dependency

2

**Is your feature request related to a problem? Please describe.** We integrate PyTorch-Tabular in [Lightning-Flash](https://github.com/PyTorchLightning/lightning-flash) thanks to the work of @Actis92 It's awesome!!! Giving us access to a great range...

Find a way to make the gpu labels more visually appealing

The highlighting of active properties is a little over the top

The border between sebmenus has double width and looks ugly when multiple menus are open

Add support for soft targets

3

## What does this PR do? Fixes # (issue) ## Before submitting - [x] Was this **discussed/approved** via a Github issue? (no need for typos and docs improvements) - [x]...

[RFC] Deprecate the `unfreeze_milestones` finetuning strategy?

6

## Motivation The `unfreeze_milestones` finetuning strategy is confusing: - how is the layer number interpretted? Does this include e.g. batch norm and non-linearity layers? Not documented - what's the use...

Add a way to track breaking changes

1

## 🚀 Feature We should have a way to track breaking changes (maybe something like the CHANGELOG) so that each release we can have the list of breaking changes in...

Finetune `SemanticSegmentation` with `ImageEmbedder` backbone

3

### Discussed in https://github.com/PyTorchLightning/lightning-flash/discussions/1288 Originally posted by **sirtris** April 11, 2022 Hi, I have pre-trained my own network backbone using the `ImageEmbedder`. I would now use this model in a...

When y_pred is a tuple (i.e. model returns multiple outputs) the criterion now recieves the tuple unpacked. This should either be reverted or more clearly documented.

Should include most of the experimented modes https://arxiv.org/abs/1805.11272