I’m following the tutorial here: SetFit: Efficient Few-Shot Learning Without Prompts
How can I access, extract, or otherwise determine the feature vector (encoded sentence via the sentence transformer) before it is fed into the final logistic regression classifier?