AlgoPuzzleVQA

该数据集旨在挑战并评估多模态语言模型在解决算法谜题方面的能力，这些谜题不仅需要视觉理解和语言理解，还需要复杂的算法推理。谜题涵盖了数学和算法的广泛主题，如布尔逻辑、组合数学、图论、优化和搜索。该数据集在推理复杂性和数据集大小上均可任意扩展。其任务包括多模态谜题解决和视觉问答。

This dataset is designed to challenge and evaluate the capabilities of multimodal language models in solving algorithmic puzzles, which require not only visual and linguistic comprehension but also complex algorithmic reasoning. The puzzles cover a wide range of mathematical and algorithmic topics, such as Boolean logic, combinatorics, graph theory, optimization, and search. This dataset can be arbitrarily scaled in terms of both reasoning complexity and dataset size. Its tasks include multimodal puzzle solving and visual question answering.