MMPR (Multimodal Preference Dataset) is a large-scale multimodal preference dataset jointly released in 2024 by research teams from Shanghai Artificial Intelligence Laboratory, Fudan University, Nanjing University, Chinese University of Hong Kong, Tsinghua University and SenseTime. The related paper results are "Enhancing the Reasoning Ability of Multimodal Large Language Models via Mixed Preference Optimization". The dataset contains 750,000 samples without clear correct answers and 2.5 million samples with clear correct answers. The samples cover multiple fields such as VQA, science, diagrams, mathematics, OCR, and documents to ensure diversity. When constructing the dataset, the researchers paid special attention to avoiding false positive negative responses due to the limitations of heuristic rules, especially in the fields of general VQA and documents. The dataset is designed to improve the performance of the model in multimodal reasoning tasks while avoiding potential negative effects during training.