Idavidrein/gpqa
Viewer
•
Updated
•
1.25k
•
51.5k
•
202
Datasets to decontaminate the post-training mixtures against. Use the subset and column values described per entry
Note Subset: gpqa_diamond Column: Question
Note Column: problem
Note Column: problem
Note Column: problem
Note Column: problem
Note Column: problem
Note Column: problem
Note Subsets: v4 & v4_v5 Column: question_content
Note Split: test Column: problem
Note Subsets: decontaminate against all 57. May be best to create a copy of the dataset and have a single `all` subset Column: question
Note Column: turn_1_prompt (needs preprocessing)
Note Subset: default Column: Question
Note Split: test