textcat.print-stream behavior does not match documentation

,
1

Version 1.5.1

This should take [dataset] [spacy_model] [source] as positional arguments, but it only seems to take [source]:

usage: prodigy textcat.print-stream [-h] [-l] [-a None] [-lo None] source

    Pretty print stream output.
    

positional arguments:
  source                Source data, file path or API query

optional arguments:
  -h, --help            show this help message and exit
  -l , --label          Label to annotate
  -a None, --api None   API loader to use
  -lo None, --loader None
                        Loader to use (if not set, loader is based on file
                        extension)
2

Thanks for the report and sorry about that! I think the reason this recipe doesn’t add labels via a model (like ner.print-stream) is that it’s more difficult to determine which ones to show. For NER, it can just highlight all entities – but if your model has 10 entries in doc.cats, it doesn’t always make sense to print all of them (and only selecting the highest score also isn’t necessarily what the user wants). Will think of a good way to solve this for the next release!

3

Thanks, that makes sense–I trust you guy’ll deliver UX-wise :slight_smile: