datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
math_shepherd
Math-Shepherd Dataset
Summary
The Math-Shepherd dataset is a processed version of Math-Shepherd dataset, designed to train models using the TRL library for stepwise supervision tasks. It provides step-by-step solutions to mathematical problems, enabling models to learn and verify each step of a solution, thereby enhancing their reasoning capabilities.
Data Structure
Format: Standard
Type: Stepwise supervision
Columns:
"pompt": The problem statement.… See the full description on the dataset page: https://huggingface.co/datasets/trl-lib/math_shepherd.math_stage2_hard_1Math-Step-DPO-10K
Step-DPO: Step-wise Preference Optimization for Long-chain Reasoning of LLMs
🖥️Code | 🤗Data | 📄Paper
This repo contains the Math-Step-DPO-10K dataset for our paper Step-DPO: Step-wise Preference Optimization for Long-chain Reasoning of LLMs, Step-DPO is a simple, effective, and data-efficient method for boosting the mathematical reasoning ability of LLMs. Notably, Step-DPO, when applied to Qwen2-72B-Instruct, achieves scores of 70.8% and 94.0% on the test sets of MATH and GSM8K… See the full description on the dataset page: https://huggingface.co/datasets/xinlai/Math-Step-DPO-10K.math_stratos_scale_judged_and_annotated_with_difficultymath-story-problems
Math Story Problems Dataset
Dataset Description
This dataset contains mathematical word problems presented in multiple formats, from direct equations to complex story-based scenarios. It is designed for training and evaluating language models on mathematical reasoning tasks.
Dataset Structure
The dataset is split into three parts:
Train: 131,072 samples
Validation: 1,024 samples
Test: 3,072 samples
Features
{
"eq_qs": "string", # Equation… See the full description on the dataset page: https://huggingface.co/datasets/azminetoushikwasi/math-story-problems.MATH_SFTmath_stackexchange_qa
Math StackExchange Curated (Parquet, CC BY-SA 4.0)
This dataset is a curated collection of Math StackExchange (MSE) Q&A pairs packaged in Parquet format.Each sample contains a problem (title, question_body), its corresponding answer (answer_body), the original MSE tag string (tags), and a flag indicating whether the answer was accepted (accepted).
This dataset includes content derived from the Math StackExchange public data dump (CC BY-SA 4.0, © Stack Exchange Inc.).This derived… See the full description on the dataset page: https://huggingface.co/datasets/glopezas/math_stackexchange_qa.sat-mathsmaths-vision-task-splitsMATH_SHEPHERD_DPO_FORMATMath-Shepherd-PRM-formatmath_stage1_8WebInstructSub-mathstackexchange-Kotiger_web_sub_mathstackexchange_templatedMath-Shepherdmaths_handwriting_imagesMMIQC-MathStExMath_syn_sourcemath_stage1_1Math-steptok-steps-mcvalue-train-part4-of-5math_stage1_10Math-Shepherdmath_stage1_11math_stage1_2Math-Shepherdmaths_bench
MathsBench
This repository contains the dataset for conducting the Large Language Model (LLM) Mathematics Benchmark.
If you find this work relevant or helpful to your work, please kindly cite it:
@misc{mathsbenchdataset,
title={MathsBench Dataset},
author={Finbarrs Oketunji},
year={2025}
}
Copyright
(c) Copyright 2025 Finbarrs Oketunji. All Rights Reserved.
math-select-06062025
Math Selection Dataset
This dataset contains 82619 examples from various math and general sources for data selection experiments. Each example contains:
text: The content, with math examples following the "Problem:\n...\n\nSolution:\n..." format
original_source: Source information for the example
Formats Available
Parquet: Available in the data/ directory
Jsonlines: Available as src.jsonl in the root directory
Sources:
TIGER-Lab/MathInstruct,train,3000:… See the full description on the dataset page: https://huggingface.co/datasets/zipfit/math-select-06062025.nyu-dl-teach-maths-comp
Math Question Answer Verification Competition Dataset
Dataset Description
The dataset for this task is the Math Question Answer Verification Competition dataset, available on Hugging Face.
It consists of questions from various math topics, structured into several columns:
question: The math question posed to the student.
answer: The ideal or correct answer to the question.
solution: A detailed reasoning or solution that explains the answer, which participants can use to… See the full description on the dataset page: https://huggingface.co/datasets/ad6398/nyu-dl-teach-maths-comp.mathstack-qamath-stratos-unverified-scaled-0.5