I'm getting error while starting coref.manual recipe
Prodigy version:- 1.10.0
Python version:- 3.6.9
Spacy version:- 2.3.0
Hi! As the error message states, the problem is that the model you're loading in doesn't have a part-of-speech tagger. The coref annotation recipe uses patterns with part-of-speech tags to pre-select and merge the relevant spans for you, including named entities, noun phrases and pronouns. Without a tagger, that's not possible, so the recipe isn't going to work,
So you can either use a pretrained model with a parser or add the default pretrained tagger to your custom model. (You could even just copy-paste the tagger directory from en_core_web_sm into your custom model and add "tagger" to the pipeline in the meta.json.) Or, if you don't want to preprocess anything and label tokens and spans from scratch, you can use the fully manual rel.manual workflow.