Skip to content

Latest commit

 

History

History
21 lines (16 loc) · 581 Bytes

TODO.md

File metadata and controls

21 lines (16 loc) · 581 Bytes
Raw

Now

  • Update convert_to_conll with hi4nlp rules

  • Support validation tokens/tags

  • Support BERT ( maybe use SrlReader)

  • Use one_hot features

  • Use POS as features to model

  • Use F1 during training/validation

  • Evaluation Framework

  • Eu treino usando o verbo do spacy, mas o original considera a tag para o verb_indicator

Testar o dataset do https://www.kaggle.com/shankkumar/multilingualopenrelations15/

Future

https://github.com/princeton-nlp/PURE Testar https://gaotianyu1350.github.io/assets/simcse/simcse.pdf

Moon-shots

  • Use GPT as a way to validate extractions