openeurollm/

smoltalk2-decontaminated

This dataset is a decontaminated version of HuggingFaceTB/smoltalk2.

MATH500: HuggingFaceH4/MATH-500 (subset=default, split=test)

AIME24: HuggingFaceH4/aime_2024 (subset=default, split=train)

AIME25: math-ai/aime25 (subset=default, split=test)

AMC23: math-ai/amc23 (subset=default, split=test)

JEEBench: daman1209arora/jeebench (subset=default, split=test)

GPQADiamond: Idavidrein/gpqa (subset=gpqa_diamond, split=train)

LiveCodeBench:… See the full description on the dataset page: https://huggingface.co/datasets/openeurollm/smoltalk2-decontaminated.