Feature Extraction with FastAI v2

I have trained a Resnet 50 model using FastAI v2. I want to know how to use the learn.model as a feature extractor and further use this embedding/vector for downstream tasks.
I believe FastAI v1 had a concept of hooks but couldn’t find any such information for FastAI v2.

Any guidance and information would help.

We have them here too :slight_smile:

2 Likes

Do we have any sample code/tutorial talking about how to use model hooks for feature extraction ?