zide05 / advfact Goto Github PK

Hi,

Thanks for the great work. In your augmented training data, e.g., in sub_ref_adv/data-train.jsonl, the first sample, the field of noise is True, which means you inject some noise to the claim by repeating some of its words, and it should be a negative sample. However, the field of label is CORRECT for this sample. I would like to ask if this sample has a problem? Thanks!

Hi,

I was trying to use the framework on a custom dataset but I am unable to figure out the correct version for the benepar dependency. The latest version supports BeneparComponent(benepar_en3) while the code uses BeneparComponent(benepar_en) which is leading to some unwanted errors while defining the nlp.pipeline ..

Would appreciate your help in this regard

Thanks!