Is there a way to edit a label (within prodigy)? So, instead of saying "no, this label is wrong" you actually go in and edit the label.
Hi! If you're using a manual interface (e.g. in recipes like
ner.make-gold), you'll be able to change the label or remove incorrect suggestions and re-add them. You can also write a custom recipe that streams in the examples, uses some objective to decide whether to ask you about a suggestion or not, and then uses a manual interface to show you the example and make it editable (e.g.
choice for text classification label or
ner_manual for named entities).
If you're using a workflow like
ner.teach, editing the examples doesn't always make sense because you're not actually giving feedback on the model's most confident predictions: you're looking at all possible analyses of the given text and are giving specific feedback on the less certain ones. So if your goal is to create a dataset of the "perfect" gold-standard annotations, it probably makes more sense to use a manual or semi-manual workflow like