This dataset contains 149960 examples and is synthetically created to enhance model's capabilities to answer complex and hard math word problems.
To generate diverse math questions, we expand the methodology in Ge et al., 2024 by using personas. More details and exact prompts used to construct the dataset can be found in our paper.… See the full description on the dataset page:
https://huggingface.co/datasets/allenai/tulu-3-sft-personas-math.
