During ner.correct annotation, something went wrong. The error message seems to be: TypeError: t[n] is undefined (as you can see on the image below). Am I doing something wrong ? Do you have an idea what it could be ?
The model used to predict annotations is a model obtained with prodigy train ner recipe (spaCy version is 2.3)
The sentences to annotate come from a jsonl file. Here are two examples:
{"text":"The presence of activated caspase-6 and cleavage of caspase-6 substrates is indeed a hallmark of AD, HD and cerebral ischemia, and has been shown in a number of different animal models and patient brain tissue [18], [19], [20], [21], [22].","meta":{"source":"sentence 180072"}}
{"text":"Caspase-6 activity can be localized to the nucleus, which is commonly associated with cell death and the cleavage of nuclear substrates such as lamin A, whereas a cytoplasmic localization of active caspase-6 as it is the case in neurodegeneration does not result in immediate apoptosis [5], [21], [32].","meta":{"source":"sentence 180124"}}
It seems like this error came out several times but not always at the same annotation. I hope this is going to help. Do not hesitate if you need further information.
Thanks! This all looks pretty standard I saw you're still on v1.10.0, though – could you try upgrading to the latest version and see if it still happens? Maybe this is related to a problem that's already fixed.