I'm using AllenNLP for SRL, but the output doesn't provide role information / frame info / lemmas for the verb. All of that would be helpful.
I see that this information seems to exist here: on the models github, but I'm not sure if it's accessible for use with the latest/best SRL models that I believe use BERT.
Would it be possible to get the sense, frame, and lemma information alongside the SRL parse, using the state-of-the-art? If so, how is it done?