I have been busy annotating data and I am trying to change the UI of prodigy to allow more text at once on the screen. The main reason that I want to do that is that I am doing a ner.correct and there are fairly little wrong annotations on a typical line / sentence and my progress is significantly slower because I need to press the enter key every second.
I added this in the prodigy.json
However, this changes the text size but doesn't add more text to my annotation window, it still shows 1 line everytime. I am using a custom pre-trained spacy model with xlm-Roberta-base which doesn't have a sentencesplitter and I wonder if that might be the reason?
Hi! The cardMaxWidth and the other theme settings really only affect the visual presentation. So making the annotation card narrower or wider won't affect the text that's displayed – this is still determined by whatever you stream in.
When you're running ner.correct, are you setting --unsegmented on the CLI? If not, the text will be split into sentences automatically. If your text comes pre-segmented and you want to stream it in as-is, you can set --unsegmented to disable this behaviour.
@ines Where do I set the --unsegmented flag? I'm getting an error on the command line when I try to add --unsegmented and -U. I have a prompt in a view_id field text box that is currently only showing a portion of the prompt text.