Update torchvision.tranforms version to torchvision.transforms.v2

[lixit]

According to https://pytorch.org/vision/stable/transforms.html#v1-or-v2-which-one-should-i-use

[HuggingFaceDocBuilderDev]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.

[tfriedel]

There's more v1 transforms in https://github.com/huggingface/pytorch-image-models/blob/45eb69d19dff83b49be14511704ec654bd24ae20/timm/data/transforms.py#L399 Maybe in other places as well, haven't checked.

Also v2 support mixup/cutmix. Maybe it's not so trivial to modify timm to use them, but I guess they will be faster.

[rwightman]

Thanks for the work but there is no point in switching v2 for the sake of it, there will be issues with timms own transforms.

I will likely keep legacy transforms as is for repeatability / bwd compat and support v2 and albumentations in cases moving forward (ie for obj detection & segmentation). I have some v2 code in the works being tested for obj detection.