- SemEval:SemEval 2010 Task 8 does not establish a default split for development, we use a random slice of the training set with 6500 examples. The data is placed in the data/new_semeval.
The processed data are placed in data/nyt and data/webnlg respectively.
- Download bert-base-uncased and bert-base-cased into the bert/ directory [google drive].