Not yet importedNo sandbox required

BoolQ

Upstream publisher Google Researchgoogle-research-datasets/boolean-questions

Naturally occurring yes-or-no questions answered from short passages, designed to test inference beyond keyword overlap.

reading-comprehensionbinarychat

PROVENANCEUPSTREAM RECORD

Upstream source: google-research-datasets/boolean-questions
Trunchbull port: Not packaged
Version: DATA / HEAD
Release digest: Not compiled
License: CC BY-SA 3.0

Source commit: Upstream head not pinnedPublished Indexed July 30, 2026

WHAT THIS PROJECT MEASURES

README

BoolQ pairs a passage with a question whose answer is true or false. The questions were written before the supporting passage was selected, which produces more natural information needs than synthetic templates.

It is a low-complexity import for Trunchbull: public text inputs, binary labels, and deterministic grading with no sandbox.

EXECUTION CONTRACT

Binary exact-match grading in the standard chat lane.

TRUNCHBULL AVAILABILITY

Import status

Not yet imported

Not yet imported. Trunchbull only needs to pin the splits, render passage-question prompts, and normalize yes/no outputs.

Port this benchmark

1 CASES

Evaluation cases

boolq/train

Passage inference

Chat lane

Answer a natural yes-or-no question using the supplied passage.

0 REQUESTED TOOLS

Tools

No model tools required.

Evaluation happens through the prompt and grader contract for this release.

MACHINE-READABLE PROVENANCE

Registry record

{
  "name": "BoolQ",
  "catalogOwner": null,
  "source": {
    "url": "https://github.com/google-research-datasets/boolean-questions",
    "version": "Upstream dataset",
    "commit": "Upstream head not pinned",
    "license": "CC BY-SA 3.0"
  },
  "port": null,
  "release": {
    "digest": "Not compiled",
    "lane": "chat",
    "sandboxRequired": false
  },
  "cases": [
    "boolq/train"
  ],
  "tools": []
}

BUILD ON THIS PROJECT