Incorrect textcat.batch-train example in documentation?

The documentation for text classification gives the following example command line.

prodigy textcat.batch-train gh_issues /tmp/model --eval-split 0.2 --label DOCUMENTATION

I think this is wrong. According to prodigy textcat.batch-train --help this recipe does not take a --label option and uses --output-model to specify the output path. I think the correct command line is

prodigy textcat.batch-train gh_issues --output-model /tmp/model --eval-split 0.2

You’re right, thanks :+1: Will fix this!

The --label used to be required, but we’ve since updated the text classifier to detect the available labels automatically from the data. (Interestingly, I noticed that using --output instead of --output-model seems to work as well – I think this might be Plac doing some magic or autocomplete behind the scenes? But anyway, we should definitely list the correct and “official” commands in the docs.)

I just ran into the same problem. The documentation still seems to be outdated