#classification-head

Timed test

1 question · 1 min · score and explanations at the end

Q 1 / 1Score 0 / 0

Shuffle questions

You fine-tune a pretrained BERT-style encoder for text classification by adding new layers. Where do they go?

On top of the last layer, fed with the output at the <cls> token1
Below the first layer, between the token embeddings and the pretrained blocks2
Inside every encoder block, replacing its pretrained feed-forward network3
Directly on the token embeddings, so that the pretrained encoder is skipped4

Keys: 1–9 select · Enter check / next