Download WoW dataset and place it under data
directory.
wget http://parl.ai/downloads/wizard_of_wikipedia/wizard_of_wikipedia.tgz
mkdir wizard_of_wiki
tar -xvf wizard_of_wikipedia.tgz -C wizard_of_wiki/
Download Wizint dataset and place it under data
too.
wget http://parl.ai/downloads/wizard_of_internet/wizard_of_internet.tgz
mkdir wizard_of_internet
tar -xvf wizard_of_internet.tgz -C wizard_of_internet/
Use our scripts to preprocess the datasets.
bash preprocess.sh
Note: nlgeval
is required for preprocessing.