prakharguptaz / Instructdial

Code for the paper Code for the paper InstructDial: Improving Zero and Few-shot Generalization in Dialogue through Instruction Tuning
Apache License 2.0
96 stars 13 forks source link

Incomplete Data Download Logic #1

Closed dzorlu closed 2 years ago

dzorlu commented 2 years ago

Hello- Thank you for releasing the code and this most comprehensive dialogue dataset.

I may be missing something, but I noticed that some of the dataset downloads seem broken. Are you going to push an updated version of the code?

For example,

dialoglue should work as following:

 git clone https://github.com/alexa/dialoglue.git
cd data_utils 
bash download_data.sh
cd ..

dialogre shd work as following:

 git clone git@github.com:nlpdata/dialogre.git

mkdir ./datasets/dialogre
wget -P ./dialogre/data_v2/en/data https://raw.githubusercontent.com/nlpdata/dialogre/master/data_v2/en/data/dev.json
wget -P ./dialogre/data_v2/en/data https://raw.githubusercontent.com/nlpdata/dialogre/master/data_v2/en/data/test.json
wget -P ./dialogre/data_v2/en/data https://raw.githubusercontent.com/nlpdata/dialogre/master/data_v2/en/data/train.json
 rm -rf dialogre/.git

Getting the following error when wow dataset download logic is run

python: can't open file './wizard_generator.py': [Errno 2] No such file or directory

There may be more, these are the issues I have run into so far. I will be also happy to push the fixes.

Thank you Deniz

prakharguptaz commented 2 years ago

Thanks for raising the issues. We are actively trying to fix them. We will push a fix in 2-3 days.

dzorlu commented 2 years ago

Hey- Just pushed a fix here. Please let me know if the PR works for you. Thanks for this awesome dataset again.

Deniz

exe1023 commented 2 years ago

Fixed the error related to WoW dataset in the commit https://github.com/prakharguptaz/Instructdial/commit/a429b361dd58964a7ccb1c3a9ceedcf304d60e36.