Skip to content

Conversation

@sunnnymskang
Copy link
Collaborator

Processing Bigbio QA dataset to Machamp classification and sequence type.
In detail,

  1. All Bigbio QA dataset has been converted to classification formatting
  • yesno qa: directly converted to classification in the following format "text": context +' ' + [SEP]+ ' '+ Question / 'labels' : yes/no
  • multiple choice qa (MCQA) in the following format: "text": context +' ' + [SEP]+ ' '+ Question + "Answer is ' + Answer / 'labels': yes/no
  • Each choice in MCQA has been converted to data points (i.e. if MCQA has 4 choices, then four classification Machamp data was created
  1. For those MCQA dataset that has exact match between answer and prompt has been converted to MaChamp sequence type
  • format: 'sequence': tokenized (context + question) /'labels': ans/none

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant