SGD pre-treatment consulting

[peng913]

Dear author, may I ask why the problem was adjusted in line 126 of the prefix_data_loader.py file schema = adjust_sgd_questions (schema).

[cuthalionn]

Hi, thanks for your question.

This is just to annotate questions for each slot that we can use to prompt the LM with, like we do for the MultiWOZ dataset. I believe I borrowed this from the TransferQA paper's data loading pipeline here: https://github.com/msamogh/Zero-Shot-DST/blob/c6a60046f166e6b5d9e46a7e7ea0c39fb0953046/TransferQA/data_loader.py#L378