Datasets are the fuel for AI. Given the rapid growth in algorithm complexity and broader adaptation of AI technologies in various domains, the requests for larger and more diverse datasets have exploded. Baidu is in a unique position to collect data from the various business units. To foster innovations in AI and expedite AI systems’ dissemination, we have decided to systematically open up datasets we have collected and used for various applications within Baidu. Our datasets are collected from real usage cases and at industrial scale and quality. Baidu Research has committed to provide these datasets at no cost for research and personal uses.