Dataset · HuggingFaceH4
30 AIME 2024 tasks for LLM evaluation
HuggingFaceH4/aime_2024
30 math competition problems from AIME 2024 to test reasoning models.
What's inside
The collection contains 30 problems from the 2024 AIME I and AIME II tests. The source is AI-MO/aimo-validation-aime, which compiles 90 AIME 2022–2024 problems. Data is text-only; the volume is under 1k examples.
What you can do with it
- Evaluating mathematical reasoning in LLMs
- Testing chain-of-thought and solver pipelines
- Comparing models on competition problems
- Size
- up to 1K rows
- Downloads this month
- 41.1K
- Likes
- 71
Models trained on it
How to load it
from datasets import load_dataset
ds = load_dataset("HuggingFaceH4/aime_2024")