Skip to content

Latest commit

 

History

History
13 lines (10 loc) · 512 Bytes

README.md

File metadata and controls

13 lines (10 loc) · 512 Bytes
Raw

qa4ie

qa4iecode (qa4ie original code)

scoringcode (scoring model code)

Model based on BIDAF

  1. Use basic/preproscr.py to preprocess the original dataset.
  2. Run basic/cli.py train the model and get inference result.
  3. Use basic/prodata.py and basic/preproqa.py to get qa4ie dataset with compressed context sentences.

scoreqacode (qa4ie with sentence selection)

Similar to qa4iecode, with data interface modified. Get new dataset through the method in scoringcode above. Run model as qa4iecode.