cff-version: 1.2.0
message: "Cite FindStatBench and the upstream FindStat database; record the exact corpus and prompt version."
title: "FindStatBench: Evaluating Large Language Models on Combinatorial Code Synthesis"
type: dataset
authors:
  - family-names: Dan
    given-names: Soham
version: 1.0.0-hf1
license: CC-BY-SA-3.0
abstract: "Lossless Hugging Face packaging of the frozen FindStatBench corpus. Data license: CC BY-SA 3.0 following the documented historical FindStat policy; software remains MIT. See DATA_RIGHTS.md for licensing scope and qualifications."