dmis-lab

/

bern2-ner

like 7

Hi, I'm trying to use this model on Sagemaker, but from the outputs I don't think I'm getting the token classification results.
I was able to set up NER inference on my personal workstation using the local installation instructions mentioned on the main BERN2 website, but unable to use the instructions here for setup on Sagemaker.
Eg -

sentence = 'Adding cataract and glaucoma as diseases so we know it works'
predictor.predict({"inputs": sentence})

[{'entity': 'I',
  'score': 0.6148244738578796,
  'index': 1,
  'word': 'Adding',
  'start': 0,
  'end': 6},
 {'entity': 'I',
  'score': 0.5745323896408081,
  'index': 3,
  'word': 'Ġand',
  'start': 16,
  'end': 19},
 {'entity': 'I',
  'score': 0.6292531490325928,
  'index': 5,
  'word': 'Ġas',
  'start': 29,
  'end': 31},
 {'entity': 'I',
  'score': 0.5479934215545654,
  'index': 8,
  'word': 'Ġwe',
  'start': 44,
  'end': 46}]```

Any input is appreciated, thanks!