I have trained a model on a certain entity. And now I would like to use the model to train another type of entity which has relationship to the one it has been trained.
I ran the trained model on a new dataset, and after just 8 annotations, it gives me a “no stream available” message. Is there any reason for this?
I’m using ner.teach recipe with input model as the trained model I had previously. And the new dataset contains 8000+ texts for annotation.
Kindly advice, thanks!
Was the entity you trained a new entity type and if so, how many examples did you use to train it?
When you see the “no tasks available” message, it typically means that Prodigy wasn’t able to find any more suggestions in the incoming stream of examples. So if your new entity type is still very rarely predicted by the model, it’s possible in theory that you only get a handful of suggestions for a given label. It definitely is surprising, though. When you run Prodigy with
PRODIGY_LOGGING=basic, is there anything in the log that looks suspicious?