single source · data curation
N-gram verbatim-overlap detection
Verbatim overlap measures shared word n-grams between a sample and a benchmark question for contamination screening.
The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See data filtering for the methods filed beside it.
Also called Verbatim overlap.
Filed under data curation › data filtering