Is it normal for ner.teach to suggest everything with SCORE: 1.00 ?

Version 1.11.7
Model is ner.manual trained in prodigy.

Hi @nsole , I might need to ask a few more clarifying questions to better help you:

  1. What model are you using? Is it something you trained yourself or what of the standard spaCy ones? (en_core_web_xx)
  2. What's the size of your dataset and its domain?

And by everything, do you mean all tokens?

New model I trained myself. Dataset has around 1500 examples.
Yes, all tokens.