search

henryhungle / NADST

Code for the paper Non-Autoregressive Dialog State Tracking (ICLR20)
MIT License
44 stars 5 forks source link

How to preprocess the dataset #9

Open zhaokanghuiasd opened 3 years ago

zhaokanghuiasd commented 3 years ago

Hi, Henry. Thanks for the great work~ But I still don't know how to preprocess the multiwoz dataset. Could you please give me some guidance? You said you would update the pre-processing scripts before :)

henryhungle commented 3 years ago

Hi @zhaokanghuiasd , the script is preprocessed in the same manner as TRADE, with additional steps to delexicalize dialogue utterances and obtain ground-truth fertilities (e.g. number of tokens in each slot value).

I will try to replicate the preprocessing code and update the repo.