Print Precision Recall in Prodigy Training Tool

Hi,

Can you make an enhancement to print Precision, Recall and F-Measure for ner.batch-train? It will help to understand how different transformations are affecting beam search training used in Prodigy differently.

Current headers are as below and I think ACCURACY means F-Measure or Precision? Is that right?

LOSS RIGHT WRONG ENTS SKIP ACCURACY

Thanks,
Sandeep

Thanks for the suggestion. I think we can at least add an option to do that; I think it makes sense.

For now you might actually be able to just modify the function in prodigy/recipes/ner.py — the source for the recipes is included within Prodigy.