mit-nlp
extract_entities returns score of 0
Expected Behavior
So, I'm using the named_entity_extractor, trained it on some data, then extracting entities from some data the model has never seen before using the extract_entities. Expecting to get back the extracted entities with their scores ranging 0-1
Current Behavior
The entities are extracted correctly, but the score is 0.
Steps to Reproduce
Train model, give it new data, get back score == 0
- Version: downloaded on 19th of July 2019
- Where did you get MITIE: Github
- Platform: 64 bit
How can I get a score in range 0-1?
P.S. While debugging I called _f.mitie_ner_get_detection_score(dets,i) on different i values and it returned non-zero scores, but these where weird as well: -5.796280346157666e-309 and 1.07329375104e-311 one of which are negative. Is this behavior typical? If so, then how should one approach this to get a normalized score in the range of 0 to 1?