#sequence-labeling

Question 1: Single choice

You train one RNN to tag every word of a sentence with its part of speech, with a loss at each position, and another RNN to classify a review's sentiment from its last output only. How are these two uses of an RNN called?

The tagger is a transducer, the sentiment model an acceptor1
The tagger is an acceptor, the sentiment model a transducer2
Both are transducers, since both read every word3
Both are bidirectional RNNs, since both use the whole sentence4

Timed test