nusnlp / neuqe Goto Github PK

I wonder you utilize the same bpe model as AAAI conv-seq2seq for the checkpoint.

TLDR; Is there a data formatting script available?

I've downloaded CLC FCE and NUCLE CoNLL-2014 datasets, but I can't figure out how to format it so I can train the predictor and estimator models. Furthermore, I can't find validation sets in the downloaded datasets. How was that chosen in your paper?

Do you get the F0.5 score file based on official evaluation code? I try to run it and it is very slow.

I wonder how to use pre-trained model to correct the error sentences, I want to use conll2014 test data ,and i want to know what i need to do to generate a file with corrected sentences