Fastai

Forthcoming breaking changes

The next version of fastai will be 2.1. It will require PyTorch 1.7, which has significant foundational changes. It should not require any code changes except for people doing sophisticated tensor subclassing work, but nonetheless we recommend testing carefully. Therefore, we recommend pinning your fastai version to `<2.1` if you are not able to fully test your fastai code when the new version comes out.

Dependencies

- pin pytorch (`<1.7`) and torchvision (`<0.8`) requirements ([2915](https://github.com/fastai/fastai/issues/2915))
- Add version pin for fastcore
- Remove version pin for sentencepiece

New Features

- added support for tb projector word embeddings ([2853](https://github.com/fastai/fastai/pull/2853)), thanks to [floleuerer](https://github.com/floleuerer)
- Added ability to have variable length draw ([2845](https://github.com/fastai/fastai/pull/2845)), thanks to [marii-moe](https://github.com/marii-moe)
- add pip upgrade cell to all notebooks, to ensure colab has current fastai version ([2843](https://github.com/fastai/fastai/issues/2843))

Bugs Squashed

- fix TabularDataLoaders inference of cont_names to keep y_names separate ([2859](https://github.com/fastai/fastai/pull/2859)), thanks to [sutt](https://github.com/sutt)

Breaking Changes

- loss functions were moved to `loss.py` ([2843](https://github.com/fastai/fastai/pull/2810))

New Features

- new callback event: `after_create` ([2842](https://github.com/fastai/fastai/issues/2842))
- This event runs after a `Learner` is constructed. It's useful for initial setup which isn't needed for every `fit`, but just once for each `Learner` (such as setting initial defaults).

- Modified XResNet to support Conv1d / Conv3d ([2744](https://github.com/fastai/fastai/pull/2744)), thanks to [floleuerer](https://github.com/floleuerer)
- Supports different input dimensions, kernel sizes and stride (added parameters ndim, ks, stride). Tested with fastai_audio and fastai time series with promising results.

Bugs Squashed

- `img_size` attribute for `TensorPoint` is not updated properly ([2799](https://github.com/fastai/fastai/pull/2799)), thanks to [IRailean](https://github.com/IRailean)

Bugs Squashed

- Undo breaking num_workers fix ([2804](https://github.com/fastai/fastai/pull/2804))
- Some users found the recent addition of `num_workers` to inference
functions was causing problems, particularly on Windows. This PR
reverts that change, until we find a more reliable way to handle
`num_workers` for inference.
- learn.tta() fails on a learner imported with load_learner() ([2764](https://github.com/fastai/fastai/issues/2764))
- learn.summary() crashes out on 2nd transfer learning ([2735](https://github.com/fastai/fastai/issues/2735))

Bugs Squashed

- Undo breaking `num_workers` fix ([2804](https://github.com/fastai/fastai/pull/2804))