Question about the situation when `absa_type==crf`

Line 442 in cf4ddc6

if self.tagger is None or self.tagger_config.absa_type == 'crf':

I wonder to know why you tranform tagger_input into self.classifier when absa_type==crf? That is, you don't use self.tagger=CRF() when "absa_type==crf".

Thank you for your attention.

Here, we use a simple linear layer to produce emission probability matrix of shape (seq_len, num_labels), which is then fed to CRF layer to calculate the sequence-level log-likelihood (see Line 476).

OK, Thanks!