Could not convert labels to ids via label_list about farm HOT 4 CLOSED

Resolved itself? : )

Anyways, your problem is a critical point. We are currently improving the error messages when dealing with custom input data. The code hidden in the parent Processor class can be sometimes confusing for people new to FARM. You can visually inspect your data processing when we log samples to the console. Or you go in debug mode and see how the data flows, especially along the functions defined here.

Normally this error comes from input data not in the right format. Formatting issues occur frequently when dealing with comma separated files and a text column often containing the separator or newlines (or any other wierd string symbols).
So please hang in there and ask questions through our github issues page here.

from farm.

Hi, sorry accidentally pressed enter before i was done typing :D I have updated my comment from earlier.

Thanks for your support so far, will look into it tomorrow again!

from farm.

Hey @amoelle
Thanks for updating the issue. Could you try renaming the "Tonalität des Artikels/Themas" column to anything but "text_classification_label". This is our internal string value, that gets assigned to label_name. If that solves the issue, we will need to add a check for that very case.

Otherwise could you verify with self.df['Tonalität des Artikels/Themas'].unique() that really all values occuring in that column are covered by the label list?

Thanks, looking forward to your reply tomorrow.

from farm.

Renaming the label column seems to work, thank you! :)

from farm.