[Docs] How can I freeze and finetune SlowFast models?

[endtheme123]

The doc issue

I am currently working with the SlowFast model for my problem requiring domain-specific data. I have prepared my data in AVA dataset format and now I need to freeze some layers, change the head of of the model, load the pretrained, and start training. However, when I checked MMaction2's doc, I began to feel it was unclear whether the head was changed yet and how can I freeze those model's layers?

Suggest a potential alternative/fix

No response