Quadra

Fixed

- Fix documentation deployment not working properly after the latest updates

Updated

- Update mkdocs related packages to the latest versions
- Reduce the amount of warnings generated by mkdocs

Updated

- Upgrade torch from 2.1.2+cu121 to 2.4.1+cu121
- Upgrade pytorch lightning to 2.4.0
- Upgrade onnxruntime-gpu to 1.20.0
- Upgrade anomalib to support the update lightning version
- Upgrade problematic packages notified by dependabot
- Deprecate python 3.9 support
- Update mypy, ruff and pylint to the latest available versions
- Apply pre-commits on all files based on the latest versions of the tools

Fixed

- Increase atol for mixed precision model export to avoid errors on efficientnet_b0 export
- Increase atol and rtol for model export tests to avoid issues on gpu testing

Fixed

- Install python3.10 in release pipelines to avoid errors with new poetry, minor pipelines improvements

Updated

- Remove gradients logging callback from default configs to avoid slowing down the training process
- Add automatic batch size calculator as default in `all` callback configuration

Updated

- Update anomalib to v0.7.0.dev143 to fix a bug introduced in the previous version that caused the training to fail if the dataset size was smaller than the batch size.

Updated

- Update anomalib to v0.7.0.dev142 that improves the memory footprint of patchcore model training.