Frédéric Branchaud-Charron

Add metric logging

7

## Summary: Add some utils to keep track of metrics over time. This is WIP, but what do you think of this API? It would be quite simple for the...

**Describe the bug** Would be cool to validate that Data augmentation is disabled when we start the active learning loop. I think PL has a similar check.

Add structure for mkdocs instead of Sphinx

1

## Summary: Makes the documentation pretty! This is WIP as I would like for everyone to have a good look before merging. This will break our website for a bit...

Use `torchmetrics` in `baal.metrics`

4

We should swap most of our metrics for `torchmetrics` this would make our code more robust. Only ECE is special I think and we should keep our own implementation.

Improve look of documentation website

12

We have several possibilities: * Move to [Mkdocs Material](https://squidfunk.github.io/mkdocs-material/) * Update our CSS In all cases, we should redo the tree structure to be more "Profesh"

It would be great if after we are done with our active learning experiment, we provide utilities to get the performance per dataset size.

Store logs on WandB

1

We would like to share experiments on Weights and Biases to share them publicly. Do we need support inside of Baal for that?

We have results on CLINC-150 where we show that Entropy is better than Random. We learned a lot since our first tutorial on HuggingFace. We should redo the expeirment.

We want to expand Baal by reaching out to different communities. We already discussed reaching out to: * MLOps * HuggingFace * MLSpace

Coresets for unsupervised active learning.

5

**Is your feature request related to a problem? Please describe.** Cold-start is a real problem in active learning, we need to label randomly a few dozens labels before starting using...