A large-scale synthetic question-answering dataset built from real-world search intent. Questions are mined from Google autocomplete suggestions, refined for clarity, and paired with short, natural-language answers.

Dataset Summary

Each example consists of a question and a corresponding answer:

Questions are mined from Google autocomplete and refined by Granite 4.1 8B. Answers are generated by LFM2.5 350M.

The questions are… See the full description on the dataset page: https://huggingface.co/datasets/ishanb3d/synthetic_qa.