
Hi there! After adding some documents to a model we started experimenting some issues in the labeling tool, you can see in the attached image how it's "recognizing" text that is not even there and assigning it to a tag. Weirdest thing is that is recognizing text from the PREVIOUS document. So those pieces of text exist in the previous doc but for some reason they come in this document.
Any thoughts why this might be happening?