Warmup schedulers in References

[datumbox]

Resolves #4281

Adds warmup on the following recipes:

• classification
• detection
• segmentation
• video classification

This PR maintains the location where we call scheduler.step() in each recipe. Segmentation and Video classification do it on the iteration level, Classification does it on the epoch level and Detection does it in a hybrid.

Though doing it on the iteration level provides more flexibility, doing so will have slight effects on the reproducibility of existing models. These effects should be minor and largely overshadowed by other differences across runs (such as the randomness of the initialization scheme). The only reason I'm not doing the switch here is because it requires extra work which I'm deferring for when we will start retraining the models using the new utils of Batteries Included.

[fmassa]

LGTM, thanks!

Just double-checking, did you run the new schedulers in a loop to compare before / after results?

[fmassa]

Created an issue to track if we can now use a stock PyTorch scheduler for this #4438

[datumbox]

@fmassa Thanks!

Yes I did. They match really close (to the 5th decimal) most of the times. For reference here are the scripts used to test this:

Segmentation:
python -m torch.distributed.launch --nproc_per_node=2 --use_env train.py --dataset coco --model lraspp_mobilenet_v3_large --epochs 7 --lr-warmup-epochs 2

Classification:
python -m torch.distributed.launch --nproc_per_node=8 --use_env train.py --model mobilenet_v3_large --data-path /datasets01/tinyimagenet/081318/ --epochs 100 --lr-scheduler cosineannealinglr --lr 0.004 --lr-warmup-epochs 10

Detection:
python -m torch.distributed.launch --nproc_per_node=8 --use_env train.py --dataset coco --model ssdlite320_mobilenet_v3_large --epochs 2 --lr-scheduler cosineannealinglr --lr 0.1 --batch-size 12

Video:
python -m torch.distributed.launch --nproc_per_node=8 --use_env train.py --data-path=/private/home/vvryniotis/kinetics400 --train-dir=train --val-dir=val --batch-size=1 --cache-dataset --epochs 10 --lr-warmup-epochs 2 --lr-milestones 4 6 8