DeepMath-103K

  🔥 News

May 8, 2025: We found that 48 samples contained hints that revealed the answers. The relevant questions have now been revised to remove the leaked answers. April 14, 2025: We release DeepMath-103K, a large-scale dataset featuring challenging, verifiable, and decontaminated math problems tailored for RL and SFT. We open source:… See the full description on the dataset page: https://huggingface.co/datasets/zwhe99/DeepMath-103K.

Ayushnangia/bucket-deepmath-99k dataset hosted on Hugging Face and contributed by the HF Datasets community

pe-nlp/DeepMath-100K-filteredv2 dataset hosted on Hugging Face and contributed by the HF Datasets community

qingy2024/DeepMath-Reformatted dataset hosted on Hugging Face and contributed by the HF Datasets community

emilbiju/RL-MATH-DeepMath dataset hosted on Hugging Face and contributed by the HF Datasets community

qingyangzhang/DeepMath-103K-formatted dataset hosted on Hugging Face and contributed by the HF Datasets community

friendshipkim/DeepMath-103K dataset hosted on Hugging Face and contributed by the HF Datasets community

DeepMath-309K_scored - with OpenDataArena Scores

This dataset is a scored version of the original zwhe99/DeepMath-103K dataset. The scoring was performed using the OpenDataArena-Tool, a comprehensive suite of automated evaluation methods for assessing instruction-following datasets. This version of the dataset includes rich, multi-dimensional scores for both the instructions (questions) and the instruction-response pairs, allowing for highly granular data analysis and selection. All… See the full description on the dataset page: https://huggingface.co/datasets/OpenDataArena/DeepMath-309K_scored.

wyhwhy/DeepMath-103K-Augmented dataset hosted on Hugging Face and contributed by the HF Datasets community

Dataset Information

To perform RLVR (e.g. GRPO) better, this dataset is a carefully filtered version of zwhe99/DeepMath-103K, resulting in 59,873 many data.

  Filtering Procedure





  Step 0. Remove 'difficulty' < 1.0 and Deduplication

At first, there are 4 questions that the 'difficulty' is less than 1.0. These 4 data points do not make sense at all. In addition to filter out those four, based on the "question" column, we do basic deduplication. As a result, the… See the full description on the dataset page: https://huggingface.co/datasets/ChuGyouk/DeepMath-Filtered-59.9K.

MultiturnRL/DeepMath-Small dataset hosted on Hugging Face and contributed by the HF Datasets community

ykarout/deepmath-small-sft dataset hosted on Hugging Face and contributed by the HF Datasets community

future7/DeepMath-Meta-Test dataset hosted on Hugging Face and contributed by the HF Datasets community

zyzshishui0627/DeepMath-103K-openai-format dataset hosted on Hugging Face and contributed by the HF Datasets community

pe-nlp/DeepMath-75K-80K-filteredv2-difficulty dataset hosted on Hugging Face and contributed by the HF Datasets community

pe-nlp/DeepMath-0.5K-filteredv2-difficulty dataset hosted on Hugging Face and contributed by the HF Datasets community

Nishi0923/DeepMath-103K-Bespoke-Filtered-Test-2k dataset hosted on Hugging Face and contributed by the HF Datasets community

pe-nlp/DeepMath-Magistral-stage1 dataset hosted on Hugging Face and contributed by the HF Datasets community

pe-nlp/DeepMath-20K-25K-filteredv2-difficulty dataset hosted on Hugging Face and contributed by the HF Datasets community

DeepMath-103K

  📖 Overview

DeepMath-103K is meticulously curated to push the boundaries of mathematical reasoning in language models. Key features include:1. Challenging Problems: DeepMath-103K has a strong focus on difficult mathematical problems (primarily Levels 5-9), significantly raising the complexity bar compared to many existing open datasets.

Difficulty… See the full description on the dataset page: https://huggingface.co/datasets/swpdd/test.