You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
generated_ood_data.zip contains questions, for example in 1hop_AndElim_random_noadj.json:
"example2": {
...,
"test_example": {
"question": "Shumpuses are grimpuses and lorpuses. Every sterpus is a lempus and a shumpus. Each lempus is a wumpus and a vumpus. Each vumpus is a numpus and an impus. Zumpuses are fruity and yumpuses. Max is an impus and a numpus. Max is a vumpus and a lempus.",
"query": "Prove: Max is a lempus.",
"chain_of_thought": [
"Max is a vumpus and a lempus.",
"Max is a lempus."
]
}
}
But note that this is the OOD version of PrOntoQA, where we only ask for proofs and not labels (i.e. true or false), and so the CoT is the answer. If you want to use data where each example has a label, you can generate examples using run_experiment.py and omitting the flag --proofs-only.
Only COT is contained in generated_ood_data.zip.
I wonder where can I find the generated QA data (including question and answer) .
The text was updated successfully, but these errors were encountered: