Hi, Do you have some stats for each aligned strategy for something like CoNLL 2003? I was wondering which of the `first`, `reduce_mean`, or `reduce_max` could be the best alignment strategy for the NER task. Many thanks.
Hi,
Do you have some stats for each aligned strategy for something like CoNLL 2003?
I was wondering which of the
first,reduce_mean, orreduce_maxcould be the best alignment strategy for the NER task.Many thanks.