How to preprocess MultiWOZ2.1?

henryhungle / NADST

Code for the paper Non-Autoregressive Dialog State Tracking (ICLR20)

MIT License

44 stars 5 forks source link

How to preprocess MultiWOZ2.1? #1

Closed SkyAndCloud closed 4 years ago

SkyAndCloud commented 4 years ago

Hi, you wrote "We use the MultiWOZ benchmark, includin both version 2.0 and 2.1. The data can be found in the folder data in the repo root folder." in Dataset section. However, there are no folder named "data". I have downloaded MultiWOZ2.1 dataset, but how to preprocess it?

henryhungle commented 4 years ago

I will update the pre-processing scripts very soon.

henryhungle commented 4 years ago

@SkyAndCloud I updated the links to download the data in the readme file.

SparkJiao commented 4 years ago

Hi, are your pre-processing scripts the same as those of TRADE ?

henryhungle commented 4 years ago

@SparkJiao Yes the pre-processing scripts are the same as in TRADE. Additional modifications are code to delexicalize dialogue utterances.