creating my own model

Following the next steps:

  1. create the training data set : python -m prodigy ner.manual ner_dataset1 en_core_web_sm c:/Users/HoulmixA/OneDrive/Bureau/texToLad.txt --label PERSON,EMA
    IL,PHONE,UNIVERSITY,INSTITUTION,TITLE,SKILLS,PROLANGUAGE,TOOLS,CITY,LANGUAGE,DATE,ORG

  2. train and save the model : python -m prodigy ner.batch-train ner_dataset1 en_core_web_sm --output model2 --eval-split 0.2

While loading: I got this :


pipeline.pyx in spacy.pipeline.Tagger.from_disk.load_tag_map() *
*
morphology.pyx in spacy.morphology.Morphology.init() *
*
attrs.pyx in spacy.attrs.intify_attrs() *
*
KeyError: 'PUNCTSIDE_FIN' *


Don't know if someone already posted something like that because I've already searched but couldn't find anything related to it. Your help please.

Hi! Did you check that both environments are running the same version of spaCy? You can run pip list to see which versions are installed. spaCy models are always compatible across patch releases, but not across minor version updates (see here). So if you're running different versions of spaCy, that might explain what's going on.

1 Like

Thank you so much for your time and answer