bug when using PIL backend in classification references

[AetelFinch]

🐛 Describe the bug

When I try to train a model using the train.py script under references/classification with the PIL backend, I encounter an error:

ValueError: backend can be 'tensor' or 'pil', but got pil

To reproduce this issue, you can write:

git clone https://github.com/pytorch/vision && cd vision
conda create -n vision_env python=3.9
conda activate vision_env
pip install torch==1.13.1 torchvision
cd references/classification/
python train.py --data-path "path-to-dataset" --test-only --backend pil

Versions

[pip3] mypy-extensions==1.0.0
[pip3] numpy==1.24.3
[pip3] torch==1.13.1
[pip3] torchvision==0.14.1
[conda] numpy 1.24.3 pypi_0 pypi
[conda] torch 1.13.1 pypi_0 pypi
[conda] torchvision 0.14.1 pypi_0 pypi