I am trying to use the token classification template to conduct NER but to use my personal taggings, i.e labling punctuation positions, product names, etc. I followed all the steps and in the seqeval session, it whos that “{personal_tags} seems not to be NE tag”. Precision, recall, accuracy, and f1 are all 1.0, which is abnormal. Would it be the problem of the way of my tagging or is it the issue of tagging names?
2 Likes
Hi! Have you solved that issue?
I encountered the same problem all metrics have the same values did you solve the problem?
Honestly I don’t quite remember how I resolved it. Maybe for we need to be careful when we use the “IOB” tagging.
i just ignored the warnings in some notebooks i saw they added " I-" to each label to avoid this problem