datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
MATH-500
Dataset Card for MATH-500
This dataset contains a subset of 500 problems from the MATH benchmark that OpenAI created in their Let's Verify Step by Step paper. See their GitHub repo for the source file: https://github.com/openai/prm800k/tree/main?tab=readme-ov-file#math-splits
math500
Dataset Card for MATH-500
This dataset contains a subset of 500 problems from the MATH benchmark that OpenAI created in their Let's Verify Step by Step paper. See their GitHub repo for the source file: https://github.com/openai/prm800k/tree/main?tab=readme-ov-file#math-splits
math_full_minus_math500
MATH (minus MATH-500)
This dataset is derived from the original MATH dataset by Hendrycks et al.
(qwedsacf/competition_math) with all problems from the MATH-500 benchmark set removed.
Construction
Source: 12,500 problems from the MATH dataset by Hendrycks et al. (qwedsacf/competition_math)
Benchmark held out: 500 problems from the MATH-500 dataset (HuggingFaceH4/MATH-500)
Matching criterion: exact match on the problem field (see… See the full description on the dataset page: https://huggingface.co/datasets/rasbt/math_full_minus_math500.MATH-500-Russian
Карточка датасета MATH-500-Russian
Перевод датасета HuggingFaceH4/MATH-500 на русский язык,
был выполнен моделью qwen2.5:32b через
скрипты EvilFreelancer/datasets-translator.
Данный набор данных содержит подмножество из 500 задач из теста MATH, который OpenAI создал для статьи Let's Verify
Step by Step и переведённых на русский язык.
Подробности в их репозиторий на GitHub.
math500-cot-experiment
MATH-500 Chain of Thought Experiment Dataset
This dataset contains the results of an experiment testing different prompting strategies (standard, chain of thought, and gibberish chain of thought) on the MATH-500 benchmark using the Llama-3.2-1B-Instruct model.
Dataset Structure
The dataset is split into three parts:
standard: Direct prompting with no reasoning steps (500 examples)
cot: Chain of thought prompting with structured reasoning (500 examples)
gibberish:… See the full description on the dataset page: https://huggingface.co/datasets/codelion/math500-cot-experiment.formal_math500
Dataset Card for Formal Problem-Solving Benchmarks
This dataset is part of the official implementation of Beyond Theorem Proving: Formulation, Framework and Benchmark for Formal Problem-Solving, accepted as an ICML 2026 Spotlight paper.
Links:
Paper: https://openreview.net/forum?id=hgMZraPlSv
Project: https://github.com/Purewhite2019/formal_problem_solving_main
Overview
The benchmark supports three evaluation settings:
Formal Problem-Solving (FPS): Given a… See the full description on the dataset page: https://huggingface.co/datasets/purewhite42/formal_math500.MATH-500
Dataset Card for MATH-500
This dataset contains a subset of 500 problems from the MATH benchmark that OpenAI created in their Let's Verify Step by Step paper. See their GitHub repo for the source file: https://github.com/openai/prm800k/tree/main?tab=readme-ov-file#math-splits
MATH-500-Overall
MATH-500-Overall
About the dataset
This dataset of only 500 examples combines mathematics, physics and logic in English with reasoning and step-by-step problem solving, the dataset was created synthetically, CoT of Qwen2.5-72B-Instruct and Llama3.3-70B-Instruct.
Brief information
Number of rows: 500
Type of dataset files: parquet
Type of dataset: text, alpaca with system prompts
Language: English
License: MIT
Structure:
math¯¯¯¯¯⌉
school-level (100 rows)… See the full description on the dataset page: https://huggingface.co/datasets/fluently-sets/MATH-500-Overall.MATH-500
Dataset Card for MATH-500
This dataset contains 12,000 training problems and 500 test problems from the MATH benchmark that OpenAI created in their Let's Verify Step by Step paper. See their GitHub repo for the source file: https://github.com/openai/prm800k/tree/main?tab=readme-ov-file#math-splits
Compared with the existing repository, this version additionally includes the training split.
math500-bon-weighted-results
MATH-500 Best-of-N Weighted Selection Results
Dataset Description
This dataset contains the results of evaluating Best-of-N weighted selection on a subset of the MATH-500 benchmark. It was created as part of a HuggingFace internship exercise exploring how test-time compute scaling with reward models can improve LLM performance on math problems.
How It Was Constructed
1. Problem Selection
Started from the HuggingFaceH4/MATH-500 dataset (500 problems)… See the full description on the dataset page: https://huggingface.co/datasets/cmpatino/math500-bon-weighted-results.math500_post-training
MATH500 Subset for Small Model Post-Training
This dataset contains a subset of 20 problems from the MATH500 dataset.
The dataset contains the following columns:
problem_id: Unique problem id that corresponds to the unique_id from the MATH500 dataset.
problem: Text describing the problem the model needs to solve.
solution: The solution generated by OpenAI available in the original dataset.
answer: The ground truth answer.
subject: Problem's subject from 7 possible values (Algebra… See the full description on the dataset page: https://huggingface.co/datasets/cmpatino/math500_post-training.MATH-500
Dataset Card for MATH-500
This dataset contains a subset of 500 problems from the MATH benchmark that OpenAI created in their Let's Verify Step by Step paper. See their GitHub repo for the source file: https://github.com/openai/prm800k/tree/main?tab=readme-ov-file#math-splits
MATH-500-gsm8k-format
MATH-500-gsm8k-format
Dataset Description
This dataset contains 500 mathematical problems from the MATH-500 benchmark, converted to GSM8K format for step-by-step reasoning.
Dataset Summary
Source: HuggingFaceH4/MATH-500
Format: GSM8K-style step-by-step solutions with inline computation annotations
Size: 500 problems
Split: Test (original MATH-500 test split)
Conversion Process
The original MATH-500 solutions (which use LaTeX notation and… See the full description on the dataset page: https://huggingface.co/datasets/albertge/MATH-500-gsm8k-format.math500-preference-pairs-fable
MATH-500 Preference Pairs (Fable-generated)
500 preference pairs covering all 500 problems of MATH-500, generated by Anthropic's Claude Fable 5 for reward-model training in a math-RLHF project (Qwen2.5-7B, PPO/GRPO on AWS EKS).
Format
{
"idx": 0,
"problem": "Convert the point $(0,3)$ ...",
"chosen": "<complete correct solution with full reasoning>",
"rejected_1": "<plausible-but-wrong solution, error mode A>",
"rejected_2": "<plausible-but-wrong solution… See the full description on the dataset page: https://huggingface.co/datasets/weivzhang/math500-preference-pairs-fable.post-training-takehome-math500-bon16
MATH-500 Best-of-16 Post-Training Take-Home Results
A 50-problem study of test-time compute, based on the Hugging Face post-training take-home challenge. Nothing here trains or modifies a model: both the generator and the reward model stay frozen, and the only variable is how a final answer is chosen from 16 sampled candidates.
Construction
Filtered MATH-500 to levels 1-3, shuffled with seed 1, and selected 50 rows.
Generated one greedy solution per problem with… See the full description on the dataset page: https://huggingface.co/datasets/augustoFranke/post-training-takehome-math500-bon16.math500-deepseek-r1-distill-qwen-1.5b
Dataset Card for "math500-deepseek-r1-distill-qwen-1.5b"
Dataset Summary
This dataset is a distilled version of the MATH500 dataset, augmented with reasoning-based responses generated by the deepseek-r1-distill-qwen-1.5b language model. The dataset is designed to evaluate and improve the mathematical reasoning capabilities of LLMs through step-by-step solutions and final answers.
Each example consists of:
The original problem statement from MATH500
The reference solution… See the full description on the dataset page: https://huggingface.co/datasets/jsm0424/math500-deepseek-r1-distill-qwen-1.5b.math500
math500
A subset of 500 mathematical problems from the MATH dataset, covering algebra, precalculus, number theory, and geometry.
Dataset Structure
This dataset is in Hugging Face datasets format. Load it with:
from datasets import load_dataset
dataset = load_dataset("Tyrion279/math500")
math-500-ptpt
Math-500-PT
Portuguese machine translation of Math-500, a benchmark of 500 challenging math problems across various topics.
Translated using Gemma-4 31B-It.
Original Dataset: https://huggingface.co/datasets/HuggingFaceH4/MATH-500
Note: This dataset is machine translated and may contain translation errors or artifacts.
This dataset is provided as part of the AMALIA project and is included in AMALIA-Bench, a comprehensive benchmark suite for evaluating large language… See the full description on the dataset page: https://huggingface.co/datasets/amalia-llm/math-500-ptpt.math500-enhanced
Math500 Enhanced Dataset
This dataset contains LLM-enhanced versions of mathematical problems with step-by-step reasoning solutions.
Dataset Statistics
Examples: 500 (500 enhanced with LLM)
Enhancement Rate: 100.0%
Data Fields
question: The mathematical problem statement
solution: LLM-enhanced step-by-step solution
original_solution: Original solution text (for reference)
answer: Final numerical answer
level: Problem difficulty level
type: Problem… See the full description on the dataset page: https://huggingface.co/datasets/rachitbansal-harvard/math500-enhanced.ru_math500
Карточка датасета MATH-500 (перевод на русский язык)
Этот датасет представляет собой перевод на русский язык поднабора из 500 задач из бенчмарка MATH, созданного OpenAI в рамках работы Let’s Verify Step by Step.
Исходные данные взяты из репозитория OpenAI:
https://github.com/openai/prm800k/tree/main#math-splits
Описание
Датасет содержит 500 математических задач с решениями, аналогичных тем, что используются для оценки моделей на способность рассуждать пошагово.
Все… See the full description on the dataset page: https://huggingface.co/datasets/AvitoTech/ru_math500.EVAL_MATH500-OT-Qwen2.5-7B-Instruct-SkillFactory-1k_rows-RL
Column Details
Column
Description
question
The question we want the model to answer
answer
The string answer
task
The name of the task the row belongs to
prompt
The prompt we will feed into the model to solve the question
model_responses
An array of strings that the model generated to answer the prompt (usually size of 4 or 34 depending on the evaluation task)
model_responses__eval_is_correct
An array aligned with model_responses containing booleans: True when… See the full description on the dataset page: https://huggingface.co/datasets/SkillFactory/EVAL_MATH500-OT-Qwen2.5-7B-Instruct-SkillFactory-1k_rows-RL.math500-deepseek-r1-distill-qwen-14b
Dataset Card for "math500-deepseek-r1-distill-qwen-14b"
Dataset Summary
This dataset is a distilled version of the MATH500 dataset, augmented with reasoning-based responses generated by the deepseek-r1-distill-qwen-14b language model. The dataset is designed to evaluate and improve the mathematical reasoning capabilities of LLMs through step-by-step solutions and final answers.
Each example consists of:
The original problem statement from MATH500
The reference solution… See the full description on the dataset page: https://huggingface.co/datasets/jsm0424/math500-deepseek-r1-distill-qwen-14b.MATH-500
Dataset Card for MATH-500
This dataset contains a subset of 500 problems from the MATH benchmark that OpenAI created in their Let's Verify Step by Step paper. See their GitHub repo for the source file: https://github.com/openai/prm800k/tree/main?tab=readme-ov-file#math-splits
MATH-500-self-rewarding使用self-rewarding方法微调的模型，在math-500上的结果
模型：qwen2.5-7b-insturct
方法：(Self-rewarding correction for mathematical reasoning)[https://arxiv.org/pdf/2502.19613]
EVAL_MATH500-OT-Qwen2.5-7B-Instruct-SkillFactory-10k_rows-RL
Column Details
Column
Description
question
The question we want the model to answer
answer
The string answer
task
The name of the task the row belongs to
prompt
The prompt we will feed into the model to solve the question
model_responses
An array of strings that the model generated to answer the prompt (usually size of 4 or 34 depending on the evaluation task)
model_responses__eval_is_correct
An array aligned with model_responses containing booleans: True when… See the full description on the dataset page: https://huggingface.co/datasets/SkillFactory/EVAL_MATH500-OT-Qwen2.5-7B-Instruct-SkillFactory-10k_rows-RL.math-500-qwen-2.5-rpm-post-training
Dataset Card
Dataset consists of 20 randomly sampled problems for level 1-3 in MATH-500 dataset dataset and solutions, scored by a reward model.
Dataset Details
Dataset Sources
The dataset is based on MATH-500 dataset. The methods used reproduce the approach from Hugging Face blogpost "Scaling Test Time Compute with Open Models"
Methods
We used two methods to generate solutions and answers:
Greedy decoding: sampling 1 solution using temperature=0… See the full description on the dataset page: https://huggingface.co/datasets/hannayukhymenko/math-500-qwen-2.5-rpm-post-training.EVAL_MATH500-OT-Qwen2.5-7B-Instruct-QwQ-10k_rows-RL
Column Details
Column
Description
question
The question we want the model to answer
answer
The string answer
task
The name of the task the row belongs to
prompt
The prompt we will feed into the model to solve the question
model_responses
An array of strings that the model generated to answer the prompt (usually size of 4 or 34 depending on the evaluation task)
model_responses__eval_is_correct
An array aligned with model_responses containing booleans: True when… See the full description on the dataset page: https://huggingface.co/datasets/SkillFactory/EVAL_MATH500-OT-Qwen2.5-7B-Instruct-QwQ-10k_rows-RL.MATH500EVAL_MATH500-OT-Qwen2.5-7B-Instruct-RL
Column Details
Column
Description
question
The question we want the model to answer
answer
The string answer
task
The name of the task the row belongs to
prompt
The prompt we will feed into the model to solve the question
model_responses
An array of strings that the model generated to answer the prompt (usually size of 4 or 34 depending on the evaluation task)
model_responses__eval_is_correct
An array aligned with model_responses containing booleans: True when… See the full description on the dataset page: https://huggingface.co/datasets/SkillFactory/EVAL_MATH500-OT-Qwen2.5-7B-Instruct-RL.EVAL_MATH500-OT-Qwen2.5-7B-Instruct-QwQ-1k_rows-RL
Column Details
Column
Description
question
The question we want the model to answer
answer
The string answer
task
The name of the task the row belongs to
prompt
The prompt we will feed into the model to solve the question
model_responses
An array of strings that the model generated to answer the prompt (usually size of 4 or 34 depending on the evaluation task)
model_responses__eval_is_correct
An array aligned with model_responses containing booleans: True when… See the full description on the dataset page: https://huggingface.co/datasets/SkillFactory/EVAL_MATH500-OT-Qwen2.5-7B-Instruct-QwQ-1k_rows-RL.