Joao Gomes

Hi @ain-soph, as discussed offline I am happy to take the review from here, and support you in the verification of the model weights that will be shipped.

I am going to launch the other training variants today and hopefully they will be ready by Monday. Regarding the difference in resolution size we can do a post training...

> I am going to launch the other training variants today and hopefully they will be ready by Monday. Regarding the difference in resolution size we can do a post...

> ``` > torchrun --nproc_per_node=4 train.py --model swin_v2_t --epochs 300 --batch-size 128 --opt adamw --lr 0.001 --weight-decay 0.05 --norm-weight-decay 0.0 --bias-weight-decay 0.0 --transformer-embedding-decay 0.0 --lr-scheduler cosineannealinglr --lr-min 0.00001 --lr-warmup-method linear...

@ain-soph thank you for your patience and great work! I managed to train all three variants and the results look good. My plan is to update this PR in the...

Trainning commands: ``` # swin_v2_t python -u run_with_submitit.py --timeout 3000 --ngpus 8 --nodes 1 --model swin_v2_t --epochs 300 \ --batch-size 128 --opt adamw --lr 0.001 --weight-decay 0.05 --norm-weight-decay 0.0 \...

In part informed by the discussions in this ticket I am proposing a new model contribution guidelines [here](https://github.com/pytorch/vision/pull/5315). Your feedback/suggestions would be very valuable.

Just for reference I can decode the video successfully on a different environment than @datumbox ``` Collecting environment information... PyTorch version: 1.12.0 Is debug build: False CUDA used to build...

@kazhang thanks for pointing out that `_transforms_video.py` is still referred internally so we need to discuss a plan for deprecating this particular file.

you also need to add `from .crestereo import *` to `torchvision/prototype/models/depth/stereo/__init__.py`