Datasetpublic

PhilipQuirke/tinyScienceQA

tinyScienceQA Pathfinder-generated 100-example tiny subset of the text-only ScienceQA test split. This is intended for fast smoke tests and default small-sample Pathfinder runs, not as an official ScienceQA benchmark replacement. Construction source dataset: tasksource/ScienceQA_text_only source split: test source rows: 2224 tiny rows: 100 seed: 20260511 selection strategy: subject_answer_quota_topic_coverage_stable_hash The selector preserves… See the full description on the dataset page: https://huggingface.co/datasets/PhilipQuirke/tinyScienceQA.

sourceHugging Facecc-by-sa-4.0updated 5mo agoView on Hugging Face

0likes22downloads

rootREADME.md

README.md58 linesDownload Raw Back to root

1---2license: cc-by-sa-4.03task_categories:4- multiple-choice5- question-answering6language:7- en8pretty_name: tinyScienceQA9source_datasets:10- tasksource/ScienceQA_text_only11---12 13# tinyScienceQA14 15Pathfinder-generated 100-example tiny subset of the text-only ScienceQA test16split. This is intended for fast smoke tests and default small-sample Pathfinder17runs, not as an official ScienceQA benchmark replacement.18 19## Construction20 21- source dataset: `tasksource/ScienceQA_text_only`22- source split: `test`23- source rows: `2224`24- tiny rows: `100`25- seed: `20260511`26- selection strategy: `subject_answer_quota_topic_coverage_stable_hash`27 28The selector preserves subject/answer-cell proportions with exact integer29quotas, ensures topic coverage where quota constraints permit, and fills the30remaining slots by a deterministic stable hash. The resulting subset covers31`25` of `25` topics.32 33## Counts34 35Subject counts:36 37```json38{39  "language science": 47,40  "natural science": 46,41  "social science": 742}43```44 45Answer counts:46 47```json48{49  "0": 44,50  "1": 43,51  "2": 9,52  "3": 453}54```55 56The corresponding Pathfinder benchmark id is `scienceqa`, and the default tiny57dataset id is `PhilipQuirke/tinyScienceQA`.58