unrecognized arguments: --seeds in textcat.teach

[aph61]

I want to train for the label INGREDIENT using the insult-classifier example. I have defined a ingredients_seed.txt, and made a terminology list using terms.teach and the en_core_web_lg model. The resulting dataset in the SQL-database is ingredients_seeds. The next step I did was

• python3 -m prodigy dataset food “all the ingredients I need for food”
• python3 -m prodigy textcat.teach ingredients en_core_web_lg RC_2015-12.bz2 --loader reddit --label INGREDIENT --seeds ingredients_seeds

This gives me “Oops, something went wrong” The dump in the shell is “prodigy textcat.teach: error: unrecognized arguments: --seeds ingredients_seed”. Dropping the “–seeds ingredients_seed” does not help

Where is my mistake/how do I solve this?

Thanks,

Andreas

[ines (Ines Montani)]

Hi! Since we made that video, the textcat.teach recipe was updated to support pattern files just like ner.teach, which gives you more flexibility than just single seed terms. See here for details:

Seeds not recognized by textcat.teach

Ah, sorry about that! If you're using the latest version of Prodigy, the --seeds argument has been replaced with a --patterns argument – see here for the available recipe arguments, and here for an example of the patterns. You'll also find more info about this in your PRODIGY_README.html. We should probably add a note about this to our text classification video! The change makes textcat.teach consistent with ner.teach, and gives you more flexibility for the seed terms. Instead of just term…

Unfortunately, YouTube doesn't let you add these comments on top of videos anymore (I guess because they were mostly used for spam?) so we could only add a note about this to the video description.