The Nemotron-RL-math-stack_overflow dataset contains mathematical problems and solutions sourced from the Stack Overflow forums. The method of extracting problems and solutions from forum posts was similar to the one used to create the OpenMathReasoning dataset, which is described in this paper. Only problems for which an answer was extracted are included in the present dataset.
This dataset is released as part of NVIDIA NeMo Gym, a framework for building… See the full description on the dataset page:
https://huggingface.co/datasets/nvidia/Nemotron-RL-math-stack_overflow.
