datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
copycolors_mcqaThis dataset consists of formatted n-way multiple choice questions, where n is in [2,10]. The task itself is simply to copy the prototypical color from the context and produce the corresponding color's answer choice letter.
The "prototypical colors" dataset instances themselves come from Memory Colors (Norland et al. 2021) and corypaik/coda (instances whose object_group is 0, indicating participants agreed on a prototypical color of that object).
Afri-MCQA
Afri-MCQA: Multimodal Cultural Question Answering for African Languages
Paper
Overview
Afri-MCQA is the first multilingual cultural question-answering benchmark covering 8k Q&A pairs across 16 African languages from 13 countries. The benchmark offers parallel English-African language Q&A pairs across text and speech modalities, entirely created by native speakers.
Supported Tasks
Visual Question Answering (VQA): Multiple-choice and open-ended QA… See the full description on the dataset page: https://huggingface.co/datasets/Atnafu/Afri-MCQA.OmniEarth-Bench_MCQ
Dataset Summary
Each example provides:
Field
Type
Description
index
int32
Row ID
query
string
Prompt that embeds both the image context and the instruction template
question
string
Human-readable question without answer options
question_type
string
"Single Choice", "Multiple Choice"
options
list[string]
letter-labelled options
answer
string
Correct letter
image
list[Image]Images for each question, range from 1 to more than 20
L1-task..L4-task
string… See the full description on the dataset page: https://huggingface.co/datasets/initiacms/OmniEarth-Bench_MCQ.LG_ConvFin_MCQ
LG ConvFinQA MCQ Dataset
Dataset Description
This dataset contains high-quality Multiple-Choice Questions (MCQs) generated from the ConvFinQA dataset for Reward Model (RM) training.
Each question has been:
Generated with 4 carefully crafted answer choices (correct/incorrect × detailed reasoning/answer only)
Verified using hybrid consistency checking:
Self-consistency: N=10 samplings with the same model
Multi-model verification: Cross-validation with 3 different models… See the full description on the dataset page: https://huggingface.co/datasets/ssunggun2/LG_ConvFin_MCQ.swedish-medical-exams-mcq-1006-json
Dataset Card for Swedish Medical Exam MCQs
Dataset Description
This dataset contains multiple-choice questions from Swedish medical exams.
Languages
The dataset is in Swedish (sv).
Dataset Structure
Each entry in the dataset contains the following fields:
question: The question
options: An array of possible answers
answer: The correct answer
language: The language of the question (always "sv" for Swedish)
country: The country of origin (always… See the full description on the dataset page: https://huggingface.co/datasets/sarafuyu/swedish-medical-exams-mcq-1006-json.ENADE_Brazilian_national_university_examination_MCQ_483medical_textbooks_mcq
Medical Textbooks MCQs Dataset
This dataset is derived from the Textbooks corpus chunks found in the MedRAG/textbooks dataset. It augments the original text snippets with synthetically generated Multiple Choice Questions (MCQs) in JSON format, suitable for fine-tuning or evaluating language models on medical MCQ generation tasks.
Dataset Details
Dataset Description
The source data consists of text snippets from the Textbooks corpus, a collection of 18 widely… See the full description on the dataset page: https://huggingface.co/datasets/asanchez75/medical_textbooks_mcq.MNLP_M2_mcqa_datasetThis dataset contains the MCQA and instruction finetuning datasets:
The messages column is used by the instruction finetuning dataset
The choices, question, context, and answer columns are used by the MCQA dataset
For the MCQA dataset (of only single answer) contains a mixture of the train, validation and test splits from this datasets as to have for training and testing:
mmlu auxiliary train we only use the stem subsets
mmlu we only use the stem subsets
ai2_arc
ScienceQA
math_qa… See the full description on the dataset page: https://huggingface.co/datasets/andresnowak/MNLP_M2_mcqa_dataset.perception_test_mcq
Perception Test MCQ Dataset
Dataset Description
This dataset contains 1000 video question-answering entries from the Perception Test dataset. Each entry includes a video and a multiple-choice question about the video content, testing various aspects of video understanding including object tracking, action recognition, and temporal reasoning.
Dataset Structure
This dataset follows the VideoFolder format with the following structure:
dataset/
├── data/
│ ├──… See the full description on the dataset page: https://huggingface.co/datasets/advaitgupta/perception_test_mcq.russian-nmo-medical-mcq
Russian NMO Medical MCQ
Choose language / Выберите язык: Русский | English
Русский
Это датасет русскоязычных медицинских тестовых вопросов НМО с вариантами ответа.
В нем есть вопросы с одним правильным вариантом и вопросы с несколькими правильными
вариантами. Датасет подготовлен так, чтобы его можно было сразу использовать для
тонкой настройки LLM, проверки качества ответов и экспериментов с медицинским QA.
Главная идея простая: дать модели вопрос, тему и варианты ответа… See the full description on the dataset page: https://huggingface.co/datasets/drkolesnikov/russian-nmo-medical-mcq.EMS-MCQA
EMS-MCQA (subset)
EMS-MCQA is a multiple-choice question answering (MCQA) dataset focused on
Emergency Medical Services (EMS) knowledge. See more on our project page.
This repo contains an open-source subset of the full dataset, provided as
JSON files.
Dataset Summary
Each record is a JSON object with:
question (str) – question stem
choices (list[str]) – options (letters like a., b., … are kept)
answer (str) or (list[str]) – correct option label ("a" | "b" | ["a"… See the full description on the dataset page: https://huggingface.co/datasets/Xueren/EMS-MCQA.marioqa-mcq
MarioQA MCQ Dataset
This dataset contains 268 samples from the MarioQA benchmark for video question answering in Mario gameplay.
Dataset Structure
Each sample contains:
file_name: Path to the video file
question: The question about the video
options: List of 4 possible answers
answer: The correct answer (a, b, c, or d)
question_type: Type of question (e.g., event-centric)
event: The event type
temporal_relationship: Temporal relationship code
Usage
from… See the full description on the dataset page: https://huggingface.co/datasets/advaitgupta/marioqa-mcq.STL-MCQA-results
STL Prompting for Zero-Shot MCQA
Results of zero-shot Multiple-Choice Question Answering (MCQA) experiments for the paper:
Dang, Q. P., Tran-Truong, P. T., Vu, D. L., Nguyen, L. S. T., Vo, Q. T. N., & Quan, T. (2026).
Enhancing large language model performance for automatic zero-shot multiple-choice question answering via single-token logit prompting.
Computers and Education: Artificial Intelligence. DOI: 10.1016/j.caeai.2026.100578
Source code:… See the full description on the dataset page: https://huggingface.co/datasets/p-storm/STL-MCQA-results.MNLP_MCQA_datasetThis MCQA dataset (of only single answer) contains a mixture of train, validation and test from this datasets (test and validation are only used for testing not for training):
mmlu auxiliary train Only the stem subset is used
mmlu Only the stem subset is used
mmlu 10 choices auxiliary train stem
ai2_arc
ScienceQA
math_qa
openbook_qa
sciq
medmcqa A 32,000 random subset (seed 42)
copycolors_mcqa
Synthetic copycolors_mcqa (4 answer choices)
This dataset is a synthetic extension of
mib-bench/copycolors_mcqa,
restricted to the 4-choice setting used in this repository.
It keeps only these counterfactual families:
answerPosition_counterfactual
randomLetter_counterfactual
answerPosition_randomLetter_counterfactual
The export uses a single train split. Each row contains one base prompt and one source row
for each of the three counterfactual types, so the dataset is balanced… See the full description on the dataset page: https://huggingface.co/datasets/jchang153/copycolors_mcqa.alba_mcq
ALBA MCQ
Multiple Choice Version of the ALBA benchmark, a Portuguese language benchmark for proficiency in pt-PT linguistic-related tasks.
For more details, see the ALBA paper.
This dataset is provided as part of the AMALIA project and is included in AMALIA-Bench, a comprehensive benchmark suite for evaluating large language models on European Portuguese.
Citation
If you use this dataset or AMALIA in your work, please cite:… See the full description on the dataset page: https://huggingface.co/datasets/amalia-llm/alba_mcq.git-history-mcq-ru
git-history-mcq-ru
805 вопросов с вариантами ответа по истории трёх открытых репозиториев
(digitable-lol/digit, digitable-lol/digitwm, digitable-lol/flang), плюс
8 672 ответа пяти моделей и 4 878 разборов этих ответов.
Вопросы на русском. Ключ каждого выведен из вывода git-команды, и сама команда
и её вывод лежат в записи — задачу можно перепроверить, не доверяя составителю.
Набор собран для одной проверки: меняют ли что-нибудь приёмы промптинга. Девять
вариантов оформления… See the full description on the dataset page: https://huggingface.co/datasets/the-homeless-god/git-history-mcq-ru.cultura-viva-pt-mcq
CulturaVivaPT
A benchmark evaluating cultural knowledge specific to Portuguese Culture.
This dataset is provided as part of the AMALIA project and is included in AMALIA-Bench, a comprehensive benchmark suite for evaluating large language models on European Portuguese.
Citation
If you use this dataset or AMALIA in your work, please cite:
@inproceedings{simplicio-etal-2026-amalia,
title = "{AMALIA}: A Fully Open Large Language Model for {E}uropean… See the full description on the dataset page: https://huggingface.co/datasets/amalia-llm/cultura-viva-pt-mcq.copycolors_mcqaThis dataset consists of a formatted version of the Memory Colors dataset (Norland et al. 2021), formatted for n-way multiple choice where n is in [2,11].
medical-terminology-mcq
Medical Terminology MCQ
742 multiple-choice items on medical terminology — word-building (root / prefix / suffix),
abbreviation expansion and disambiguation, minimal-pair discrimination, and compositional
generalization to novel compounds.
Labels are what GPT-5.6-sol ruled they are. Items were written by Claude and adjudicated
by GPT-5.6-sol; where the two disagreed, the adjudicator's ruling settled the item. That
makes GPT-5.6-sol the source of truth for every answer key here.… See the full description on the dataset page: https://huggingface.co/datasets/chongpangnasilemak/medical-terminology-mcq.icd10cm-exam-mcq
ICD-10-CM Exam MCQ
203 multiple-choice items drawn from ICD-10-CM practice exams and coursework — clinical
vignettes requiring an actual code assignment, plus questions on conventions, guideline
structure and Chapter 20 external-cause rules.
⚠️ Read this before using or redistributing
The questions are third-party material of unverified provenance, reproduced verbatim.
They come from practice exams and coursework — one identifies itself as belonging to an… See the full description on the dataset page: https://huggingface.co/datasets/chongpangnasilemak/icd10cm-exam-mcq.multi-subject-mcq-training-pool
Multi-subject multiple-choice training pool
Public multiple-choice questions from eight datasets covering medicine, law, quantitative
reasoning, the natural sciences, history, philosophy, business, economics and everyday knowledge,
read at the pinned revisions named below and laid out twice. Train on either layer or on both.
pool.jsonl
Every source rewritten into one shape, 347642 rows, one JSON object per line, with these fields.
Field
What it holds
id… See the full description on the dataset page: https://huggingface.co/datasets/Emulated-Inc/multi-subject-mcq-training-pool.mcqa-single-answer
Dataset Summary
EVE-mcqa-single-answer is a Multiple-Choice Question Answering (MCQA) dataset designed to evaluate the performance of language models in the domain of Earth Observation (EO). The dataset consists of questions related to EO concepts, technologies, and applications, each accompanied by multiple answer choices with exactly one correct answer.
Unlike multi-answer MCQA datasets, each question in this dataset has only a single correct choice, making it suitable for… See the full description on the dataset page: https://huggingface.co/datasets/eve-esa/mcqa-single-answer.MRI-MCQA
MRI-MCQA
Dataset Description
MRI-MCQA is a benchmark composed by multiple-choice questions related to Magnetic Resonance Imaging (MRI). We use this dataset to evaluate the level of knowledge of various LLMs about the MRI field.
Curated by: Oscar Molina Sedano
Language(s) (NLP): English
License
This dataset is licensed under CC-BY-NC 4.0.
Disclaimer
Courtesy of Allen D. Elster… See the full description on the dataset page: https://huggingface.co/datasets/HPAI-BSC/MRI-MCQA.aplikacje-prawnicze-mcq
Polish Legal Apprenticeship Entrance Exams — adwokacka/radcowska, notarialna, komornicza (2007–2025)
Native-Polish, single-choice (A/B/C) legal MCQ benchmark built from the official entrance
examinations for the Polish legal apprenticeships, published by the Ministry of Justice:
adwokacka + radcowska (advocate + legal counsel — a single shared test from 2009 on;
two separate exams in 2007),
notarialna (notary),
komornicza (court-enforcement officer / bailiff).
Each item… See the full description on the dataset page: https://huggingface.co/datasets/bartoszkobylinski1/aplikacje-prawnicze-mcq.science-mcqa-training-pool
Science multiple-choice training pool
Public multiple-choice science questions from three datasets, read at the pinned revisions named
below and laid out twice. Train on either layer or on both.
pool.jsonl
Every source rewritten into one shape, 182035 rows, one JSON object per line, with these fields.
Field
What it holds
id
a row identifier unique within this file
question
the question text, as its source publishes it
options
the answer options, as… See the full description on the dataset page: https://huggingface.co/datasets/Emulated-Inc/science-mcqa-training-pool.mongolian-mcq-dataset
Mongolian MCQ Dataset with Sources
This dataset contains Mongolian multiple-choice questions across school and general-knowledge subjects. Each row includes answer choices, the correct answer, an explanation, and source metadata.
Dataset contents
File
Rows
mongolian_ap_chemistry_mcq_100.jsonl
100
mongolian_ap_physics_slightly_harder_mcq_100.jsonl
100
mongolian_biology_highschool_wikibooks_mcq_100.jsonl
100… See the full description on the dataset page: https://huggingface.co/datasets/Asakuu/mongolian-mcq-dataset.icd10pcs-coding-mcq
ICD-10-PCS Coding MCQ
405 multiple-choice items on ICD-10-PCS inpatient procedure coding — whether a model can
build a seven-character procedure code from documentation it is handed: root operation
selection, the seven character axes, Index→Tables verification, approach, device and qualifier
values, and the Official Guidelines.
Labels are what GPT-5.6-sol ruled they are. Items were written by Claude and adjudicated
by GPT-5.6-sol; where the two disagreed, the adjudicator's… See the full description on the dataset page: https://huggingface.co/datasets/chongpangnasilemak/icd10pcs-coding-mcq.medical-mcq-training-pool
Medical multiple-choice training pool
Public multiple-choice medical questions from three repositories, read at the pinned revisions
named below and laid out twice. Train on either layer or on both.
pool.jsonl
Every source rewritten into one shape, 219695 rows, one JSON object per line, with these fields.
Field
What it holds
id
a row identifier unique within this file
question
the question text, as its source publishes it
options
the answer options… See the full description on the dataset page: https://huggingface.co/datasets/Emulated-Inc/medical-mcq-training-pool.icd10cm-coding-mcq
ICD-10-CM Coding MCQ
403 multiple-choice items on ICD-10-CM diagnosis coding — whether a model can apply the
classification's conventions to documentation it is handed: Excludes1 and Excludes2 notes,
7th-character selection, placeholder X, laterality, Index→Tabular verification, combination
codes, specificity.
Labels are what GPT-5.6-sol ruled they are. Items were written by Claude and adjudicated
by GPT-5.6-sol; where the two disagreed, the adjudicator's ruling settled the… See the full description on the dataset page: https://huggingface.co/datasets/chongpangnasilemak/icd10cm-coding-mcq.