datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
MathReal
Dataset Card for MathReal
Dataset Description
Paper Information
Dataset Examples
Leaderboard
Citation
Dataset Description
The MathReal dataset is designed to evaluate the performance of Multi-modal Large Language Models (MLLMs)on real-world K-12 mathematical questions. It consists of 2,000 high-quality math problems, each represented as an image captured in authentic educational contexts. The dataset includes various types of questions, such as multiple-choice… See the full description on the dataset page: https://huggingface.co/datasets/junfeng0288/MathReal.TemplateGSM
TemplateMath: Template-based Data Generation (TDG)
This is the official repository for the paper "Training and Evaluating Language Models with Template-based Data Generation", published at the ICLR 2025 DATA-FM Workshop.
Our work introduces Template-based Data Generation (TDG), a scalable paradigm to address the critical data bottleneck in training LLMs for complex reasoning tasks. We use TDG to create TemplateGSM, a massive dataset designed to unlock the next level of… See the full description on the dataset page: https://huggingface.co/datasets/math-ai/TemplateGSM.Math-RL-Tasks
Ulam AI Math RL Tasks
Forty original, verifier-backed mathematical reasoning tasks packaged as ten
independent RL environments. The collection spans advanced graduate exercises,
research-style exact computation and structural generalization problems in
algebraic geometry, arithmetic geometry, combinatorics, topology, probability
and spectral analysis.
Each suite pairs a runnable rl_env/ with a preserved blind_run/ by
GPT-5.6 Sol Pro. The model name describes the evaluation actor… See the full description on the dataset page: https://huggingface.co/datasets/ulamai/Math-RL-Tasks.ODA-Math-460k
ODA-Math-460k
ODA-Math-460k is a large-scale math reasoning dataset curated from top-performing open mathematics corpora (selected via the OpenDataArena leaderboard) and refined through deduplication, benchmark decontamination, LLM-based filtering, and verifier-backed response distillation.It targets a “learnable but challenging” difficulty band: non-trivial for smaller models yet solvable by stronger reasoning models.
🧠 Dataset Summary
Domain: Mathematics… See the full description on the dataset page: https://huggingface.co/datasets/OpenDataArena/ODA-Math-460k.MathLake
MathLake: A Large-Scale Mathematics Dataset
MathLake is a massive collection of 8.3 million mathematical problems aggregated from over 50 open-source datasets. Unlike datasets focused on filtering for the highest quality solutions immediately, MathLake prioritizes query comprehensiveness, serving as a universal "raw ore" for researchers to curate, distill, or annotate further. The dataset provides annotations for Difficulty, Format, and Subject, covering diverse mathematical fields… See the full description on the dataset page: https://huggingface.co/datasets/OpenDataArena/MathLake.olympiad_style_integer_math_problems
Olympiad Math Corpus
Version: v2.1.1
Release date: 2026-05-03
59,486 synthetically generated olympiad-style math problems with verified integer answers and formal computation graphs.
Loading
from datasets import load_dataset
ds = load_dataset("mihailgribov/olympiad_style_integer_math_problems", split="train")
lemma_applicability is stored as list[{lemma, status}] rather than a sparse dict (required for Arrow-based consumers). To convert to a dict for local use:… See the full description on the dataset page: https://huggingface.co/datasets/mihailgribov/olympiad_style_integer_math_problems.theorem-search-dataset
Theorem Search Dataset
The largest open corpus of informal mathematical theorems: 1,341,083 theorem statements with natural-language slogans from 209,777 papers, designed for semantic theorem retrieval.
Paper: Semantic Search over 9 Million Mathematical Theorems
Demo: huggingface.co/spaces/uw-math-ai/theorem-search
Benchmark results
On 110 test queries written by research mathematicians, our best pipeline (Qwen3-Embedding-8B on DeepSeek-V3.1 slogans) outperforms… See the full description on the dataset page: https://huggingface.co/datasets/uw-math-ai/theorem-search-dataset.SAND-MATH
SAND-Math: A Synthetic Dataset of Difficult Problems to Elevate LLM Math Performance
📃 Paper | 🤗 Dataset
SAND-Math (Synthetic Augmented Novel and Difficult Mathematics) is a high-quality, high-difficulty dataset of mathematics problems and solutions. It is generated using a novel pipeline that addresses the critical bottleneck of scarce, high-difficulty training data for mathematical Large Language Models (LLMs).
Key Features
Novel Problem Generation: Problems are… See the full description on the dataset page: https://huggingface.co/datasets/amd/SAND-MATH.olympiad_style_integer_math_reasoning
Olympiad Math Reasoning Traces
Version: v1.0.2
Release date: 2026-04-19
64,763 full model reasoning traces for olympiad-style math problems with verified integer answers. This dataset contains only correct and non-truncated traces — every record contains a terminal \boxed{...} answer (within the last 500 characters of the response) that matches the expected integer exactly, and none of the responses hit the model's generation-token cap. Intended for distillation and supervised… See the full description on the dataset page: https://huggingface.co/datasets/mihailgribov/olympiad_style_integer_math_reasoning.deepseek-v4-pro-math-cot-1k
DeepSeek V4 Pro Math CoT 1K
A small, high-signal supervised-fine-tuning (SFT) dataset of math reasoning traces. Problems were sampled from a Nemotron math problem set (originally sourced from StackExchange-Math and AoPS), answered by DeepSeek V4 Pro with thinking enabled at high reasoning effort, then independently reviewed by DeepSeek V4 Flash for correctness against the expected answer. Pathological reasoning traces (looping, run-away length, excessive Wait-style backtracking)… See the full description on the dataset page: https://huggingface.co/datasets/blythet/deepseek-v4-pro-math-cot-1k.MathFusionQAPaper: MathFusion: Enhancing Mathematic Problem-solving of LLM through Instruction Fusion
Code: https://github.com/QizhiPei/MathFusion
gretel-math-gsm8k-v0
gretelai/gsm8k-synthetic-diverse-405b
This dataset is a synthetically generated version inspired by the GSM8K https://huggingface.co/datasets/openai/gsm8k dataset, created entirely using Gretel Navigator with meta-llama/Meta-Llama-3.1-405B as the agent LLM. It contains ~1500 Grade School-level math word problems with step-by-step solutions, focusing on age group, difficulty, and domain diversity.
Key Features:
Synthetically Generated: Math problems created using Gretel… See the full description on the dataset page: https://huggingface.co/datasets/gretelai/gretel-math-gsm8k-v0.oellm-math-rlvr
OpenEuroLLM Math RLVR
One million deterministic, verifier-ready mathematical problems for reinforcement learning with
verifiable rewards. The release contains a 760,000-row English depth pool and 10,000 aligned semantic
problems rendered in all 24 official EU languages (240,000 rows).
This is a prompt-and-answer rollout corpus, not a chain-of-thought corpus. Model inputs contain only the
problem and output-format instruction. Reference answers and verifier contracts remain… See the full description on the dataset page: https://huggingface.co/datasets/openeurollm/oellm-math-rlvr.math_benchmark_test_saturation
LLM Leaderboard Data for Hendrycks MATH Dataset (2022–2024)
This dataset aggregates yearly performance (2022–2024) of large language models (LLMs) on the Hendrycks MATH benchmark. It is specifically compiled to explore performance evolution, benchmark saturation, parameter scaling trends, and evaluation metrics of foundation models solving complex math word problems.
Original source data: Math Word Problem Solving on MATH (Papers with Code)
About Hendrycks' MATH… See the full description on the dataset page: https://huggingface.co/datasets/nlile/math_benchmark_test_saturation.adaption-math-worked-solutions-raw
NuminaMath Worked Solutions
Competition and school maths problems with step-by-step solutions.
Rows
3,000
Domain
mathematics
Format
data.parquet, one row per example
Licence
apache-2.0
Built for
supervised fine-tuning (SFT) experiments on Adaption AutoScientist
Columns
Column
Description
original_prompt
The prompt (user turn) as uploaded.
original_completion
The target response as uploaded.
enhanced_prompt
Empty in this… See the full description on the dataset page: https://huggingface.co/datasets/rodriguescarson/adaption-math-worked-solutions-raw.oellm-math-rlvr
OpenEuroLLM Math RLVR
One million deterministic, verifier-ready mathematical problems for reinforcement learning with
verifiable rewards. The release contains a 760,000-row English depth pool and 10,000 aligned semantic
problems rendered in all 24 official EU languages (240,000 rows).
This is a prompt-and-answer rollout corpus, not a chain-of-thought corpus. Model inputs contain only the
problem and output-format instruction. Reference answers and verifier contracts remain… See the full description on the dataset page: https://huggingface.co/datasets/birgermoell/oellm-math-rlvr.adaption-math-worked-solutions-mixtral-raw
NuminaMath Worked Solutions
Competition and school maths problems with step-by-step solutions.
Rows
3,000
Domain
mathematics
Format
data.parquet, one row per example
Licence
apache-2.0
Built for
supervised fine-tuning (SFT) experiments on Adaption AutoScientist
Columns
Column
Description
original_prompt
The prompt (user turn) as uploaded.
original_completion
The target response as uploaded.
enhanced_prompt
Empty in this… See the full description on the dataset page: https://huggingface.co/datasets/rodriguescarson/adaption-math-worked-solutions-mixtral-raw.MATH-Beyond
MATH-Beyond: A Benchmark for RL to Expand Beyond the Base Model
A benchmark dataset for evaluating reinforcement learning methods on challenging mathematical problems that push beyond base model capabilities.
Paper
This dataset accompanies our paper: MATH-Beyond: A Benchmark for RL to Expand Beyond the Base Model
Dataset Overview
MATH-Beyond (MATH-B) contains 181 carefully selected mathematical problems deliberately constructed to defeat common open-source… See the full description on the dataset page: https://huggingface.co/datasets/brendel-group/MATH-Beyond.adaption-math-numina-worked-24k
NuminaMath Worked Solutions
Competition and school maths problems with step-by-step solutions.
Rows
24,000
Domain
mathematics
Format
data.parquet, one row per example
Licence
apache-2.0
Built for
supervised fine-tuning (SFT) experiments on Adaption AutoScientist
Columns
Column
Description
original_prompt
The prompt (user turn) as uploaded.
original_completion
The target response as uploaded.
enhanced_prompt
Empty in this… See the full description on the dataset page: https://huggingface.co/datasets/rodriguescarson/adaption-math-numina-worked-24k.adaption-math-numina-worked-41k
NuminaMath Worked Solutions
Competition and school maths problems with step-by-step solutions.
Rows
41,100
Domain
mathematics
Format
data.parquet, one row per example
Licence
apache-2.0
Built for
supervised fine-tuning (SFT) experiments on Adaption AutoScientist
Columns
Column
Description
original_prompt
The prompt (user turn) as uploaded.
original_completion
The target response as uploaded.
enhanced_prompt
Empty in this… See the full description on the dataset page: https://huggingface.co/datasets/rodriguescarson/adaption-math-numina-worked-41k.Math_CoT_Arabic_English_Reasoning
Math CoT Arabic English Dataset
A high-quality, bilingual (English & Arabic) dataset for Chain-of-Thought (COT) reasoning in mathematics and related disciplines, developed by Miscovery AI.
Overview
Math-COT is a unique dataset designed to facilitate and benchmark the development of chain-of-thought reasoning capabilities in language models across mathematical domains. With meticulously crafted examples, explicit reasoning steps, and bilingual support, this dataset offers… See the full description on the dataset page: https://huggingface.co/datasets/miscovery/Math_CoT_Arabic_English_Reasoning.math-formulas
Math Formulas QA
Deterministic synthetic math QA dataset generated with seed 1337.
Properties
2,000,000 unique rows
1,800,000 train
100,000 validation
100,000 test
100,000 rows per Parquet shard
Every row is validated before it is written
kind alternates between problem and solution
Columns
question
answer
text
question_tex
answer_tex
family
difficulty
kind
validated
validator
formula_hash64
Families
Arithmetic, fractions… See the full description on the dataset page: https://huggingface.co/datasets/aplominski/math-formulas.pt-exams-math-open
PT exams, mathematics, open answer
The 166 Mathematics A questions of PHEB
(Tavares et al., LREC 2026), the Portuguese national secondary-school exams
from 2006 to 2023, converted by hand from multiple choice to open-answer form.
The set was built for the paper Does CoT-Pass@k Really Check the CoT? A
Multilingual Mathematical Audit (Taşaltı, Hüdaverdi, Semedo; MRL Workshop,
EMNLP 2026), where it is the pt_exams_math benchmark. Two graduate students,
native speakers of Portuguese… See the full description on the dataset page: https://huggingface.co/datasets/tariktuna/pt-exams-math-open.unsolved-math-clean
🧠 Unsolved Math — Clean
8,626 curated open research problems in mathematics and CS — including 122 Millennium Prize Problems — deduplicated, schema-flattened, and packaged as proper parquet configs with an eval-only benchmark view.
A reasoning frontier dataset: every problem here is actually unsolved or partially solved — ideal for honest capability probing instead of contaminated benchmarks.
Clean derivative of ulamai/UnsolvedMath (8,785 problems). License unchanged:… See the full description on the dataset page: https://huggingface.co/datasets/saidutta69/unsolved-math-clean.math_onetjapanese-math-empirical-difficulty-pilot-50k
Japanese Math Empirical Difficulty Pilot 50k
This dataset is a 50,000-problem empirical difficulty pilot, not a full empirical labeling of the original 5.66M-row source dataset.
It was created for LLM-jp experiment 0399, Team Victory SFT, to validate empirical difficulty label distribution, downstream split behavior, and the rollout/scoring pipeline before attempting labeling at the full 5.6M scale.
Current Status
This upload uses the v3 scorer with assistant-only… See the full description on the dataset page: https://huggingface.co/datasets/argo11/japanese-math-empirical-difficulty-pilot-50k.DeepSeek-v3.1-reasoner-Distilled-math-samples
DeepSeek-V3.1 Distillation with NVIDIA Nemotron-Post-Training-Dataset-v2 (Math Subset)
The release of DeepSeek-V3.1 has attracted wide attention in the AI community. Its significant improvements in reasoning ability provide a new opportunity to explore optimization of domain-specific models. To investigate the potential of this model in complex mathematical reasoning tasks, I selected the math subset from NVIDIA’s newly released Nemotron-Post-Training-Dataset-v2 as seed problems and… See the full description on the dataset page: https://huggingface.co/datasets/Jackrong/DeepSeek-v3.1-reasoner-Distilled-math-samples.OpenVerification1_aux_mathnet
Dataset Card for ReexpressAI/OpenVerification1_aux_mathnet
This is additional data as part of ReexpressAI/OpenVerification1. The data fields are slightly different for this data source, so we include this as a separate dataset.
This provides hard negatives and verifications for a subset of MathNet.
Refer to ReexpressAI/OpenVerification1 for additional context and details.
Dataset Structure
Here, we provide additional details for each field. The prompts used for… See the full description on the dataset page: https://huggingface.co/datasets/ReexpressAI/OpenVerification1_aux_mathnet.ktt-math-tutor-data
KTT Math Tutor — Data
Data artefacts for the AIMS KTT Hackathon Tier-3 submission
S2.T3.1 AI Math Tutor for Early Learners. Source code:
https://github.com/DrUkachi/ktt-math-tutor.
Contents
T3.1_Math_Tutor/
Core curriculum + seeds.
curriculum.json — 80 items × 5 sub-skills (counting, number
sense, addition, subtraction, word problem) with EN / FR / KIN
stems, difficulty 1–10, age bands 5–6 / 6–7 / 7–8 / 8–9, visual
asset keys, expected integer answer.… See the full description on the dataset page: https://huggingface.co/datasets/DrUkachi/ktt-math-tutor-data.dumy-zno-ukrainian-math-history-geo-r1-o1
DUMY («Думи»): Ukrainian Multidomain Reasoning Dataset (Part 1: ZNO/NMT tasks with DeepSeek R1 and OpenAI o1 answers)
DUMY is an open benchmark and dataset designed for training, distillation, and evaluation of language models focused on Ukrainian reasoning tasks.
The word “Dumy” comes from Taras Shevchenko’s famous poem and literally means “thoughts” in Ukrainian:
Думи мої, думи мої,
Лихо мені з вами!
Нащо стали на папері
Сумними рядами?..
Work in progress. Stay tuned.… See the full description on the dataset page: https://huggingface.co/datasets/NLPForUA/dumy-zno-ukrainian-math-history-geo-r1-o1.