Datasets:

lmms-lab
/

imo-2025

dataset_info:
  features:
    - name: url
      dtype: string
    - name: problem
      dtype: string
    - name: answer
      dtype: string
  splits:
    - name: train
      num_examples: 6
license: mit
language:
  - en
tags:
  - mathematics
  - olympiad
  - problem-solving
  - competition-math

IMO 2025 Problems Dataset

This dataset contains the 6 problems from the 2025 International Mathematical Olympiad (IMO). The problems are formatted with proper LaTeX notation for mathematical expressions. WARNING: Somehow, only Problem 1 is actually from IMO 2025. The others are from some other place!

Dataset Structure

Each example contains:

url: Problem identifier (e.g., "2025-imo-p1")
problem: The problem statement with LaTeX mathematical notation
answer: The solution (currently set to "N/A")

Problem Types

The dataset includes problems covering various mathematical areas:

Problem 1: Combinatorial geometry (sunny lines)
Problem 2: Functional equations
Problem 3: Number theory (perfect squares)
Problem 4: Combinatorial game theory
Problem 5: Number theory (primitive sets)
Problem 6: Euclidean geometry

Mathematical Notation

Mathematical expressions are formatted using LaTeX:

Variables and expressions: $x$ , $n \geq 3$
Display equations: $$f(x^3 + y^3) = f(x + y)(x^2 - xy + y^2)$$
Sets: $\mathbb{R}$ , $S$
Special formatting: sunny, primitive

Files

imo_2025.json: Full dataset in JSON format
imo_2025.jsonl: Dataset in JSONL format (one problem per line)
README.md: This file

Usage

from datasets import load_dataset

# Load the dataset
dataset = load_dataset("lmms-lab/imo-2025")

# Access individual problems
for problem in dataset['train']:
    print(f"Problem: {problem['url']}")
    print(f"Statement: {problem['problem']}")
    print()

Or load directly from JSON/JSONL:

import json

# Load from JSON
with open("imo_2025.json", "r") as f:
    problems = json.load(f)

# Or load from JSONL
problems = []
with open("imo_2025.jsonl", "r") as f:
    for line in f:
        problems.append(json.loads(line))

Citation

If you use this dataset in your research, please cite:

@dataset{imo2025,
  title={IMO 2025 Problems Dataset},
  author={LMMS Lab},
  year={2025},
  url={https://huggingface.co/datasets/lmms-lab/imo-2025}
}

Source

Problems are from the 2025 International Mathematical Olympiad. Original source: https://artofproblemsolving.com/wiki/index.php/2025_IMO_Problems

License

This dataset is released under the MIT License.