longxudou/HIT-SCIR-CoNLL2019

Issues

• prediction for new texts broken - `_json_to_instance` uses dataset_reader

  #13 opened 4 years ago by kleinay
  0

• A bug in the EDS predictor?

  #7 opened 4 years ago by ZhaofengWu
  5

• Train/dev split

  #12 opened 4 years ago by ZhaofengWu
  2

• Let's push the boundaries of the state of the art once again!

  #10 opened 4 years ago by LifeIsStrange
  4

• Evaluation metric in Table 2 in the paper

  #11 opened 4 years ago by danielhers
  2

• AMR preprocessing fails

  #2 opened 4 years ago by danielhers
  4

• Configuration file with glove embeddings

  #8 opened 4 years ago by mdelhoneux
  2

• Long sequences breaking `_greedy_decode`

  #9 opened 4 years ago by ZhaofengWu
  0

• Does this only use tokenization from the companion file?

  #6 opened 4 years ago by ZhaofengWu
  4

• KeyError when Evaluation on UCCA parser

  #3 opened 4 years ago by ccclyu
  5

• Why do different frameworks use different flavors of BERT?

  #4 opened 4 years ago by ZhaofengWu
  2

• No "anchors" key in evaluation UDPipe file

  #5 opened 4 years ago by ZhaofengWu
  2

• Pre-trained models

  #1 opened 5 years ago by danielhers
  2