Large vision language models (LVLMs) have improved the document understanding capabilities remarkably, enabling the handling of complex document elements, longer contexts, and a wider range of tasks. However, existing document understanding benchmarks have been limited to handling only a small number of pages and fail to provide a comprehensive analysis of layout elements locating. In this paper, we first define three primary task categories: Long Document Understanding, numerical Reasoning, and cross-element Locating, and then propose a comprehensive benchmark—LongDocURL—integrating above three primary tasks and comprising 20 sub-tasks categorized based on different primary tasks and answer evidences. Furthermore, we develop a semi-automated construction pipeline and collect 2,325 high-quality question-answering pairs, covering more than 33,000 pages of documents, significantly outperforming existing benchmarks. Subsequently, we conduct comprehensive evaluation experiments on both open-source and closed- source models across 26 different configurations, revealing critical performance gaps in this field. The code and data: https://github.com/dengc2023/LongDocURL.

Instruction tuning is widely used to enhance a pre-trained Multimodal Large Language Model (MLLM) to understand and follow human instructions by training it on a curated set of task-specific dataset. However, it is infeasible to collect all possible instruction datasets simultaneously in real-world scenarios. Thus, enabling MLLM with continual instruction tuning is essential for maintaining their adaptability. However, existing methods often trade off memory efficiency for performance gains, significantly compromising overall efficiency. In this paper, we propose a task-specific expansion and task-general fusion framework based on the variations in Centered Kernel Alignment (CKA) similarity across different model layers when trained on diverse datasets. Furthermore, we analyze the information leakage present in the existing benchmark and propose a new and more challenging benchmark to rationally evaluate the performance of different methods. Comprehensive experiments showcase a significant performance improvement of our method compared to existing state-of-the-art methods. Our code will be public available.

With the rapid advancements in multimodal large language models, evaluating their multimodal mathematical capabilities continues to receive wide attention. Although datasets such as MathVista have been introduced for evaluating mathematical capabilities in multimodal scenarios, there remains a lack of evaluation tools and datasets tailored for fine-grained assessment in Chinese K12 education. To systematically evaluate the ability of multimodal large models to solve Chinese multimodal mathematical problems, we propose a Chinese Multi-modal Math Skill Evaluation Benchmark (CMMaTH), containing 23,856 multimodal K12 math related questions, making it the largest Chinese multimodal mathematical problem benchmark to date. CMMaTH includes questions ranging from elementary to high school levels, offering greater diversity in problem types, solution goals, visual elements, detailed knowledge points, and standard solution annotations. To facilitate stable, fast, and cost-free model evaluation, we have developed an open-source tool called GradeGPT, which is integrated with the CMMaTH dataset. Our data and code are available at https://github.com/zzli2022/CMMaTH.

Recent advancements in large language models (LLMs) and multi-modal models (MMs) have demonstrated their remarkable capabilities in problem-solving. Yet, their proficiency in tackling geometry math problems, which necessitates an integrated understanding of both textual and visual information, has not been thoroughly evaluated. To address this gap, we introduce the GeoEval benchmark, a comprehensive collection that includes a main subset of 2,000 problems, a 750 problems subset focusing on backward reasoning, an augmented sub- set of 2,000 problems, and a hard subset of 300 problems. This benchmark facilitates a deeper investigation into the performance of LLMs and MMs in solving geometry math problems. Our evaluation of ten LLMs and MMs across these varied subsets reveals that the WizardMath model excels, achieving a 55.67% accuracy rate on the main subset but only a 6.00% accuracy on the hard subset. This highlights the critical need for testing models against datasets on which they have not been pre-trained. Additionally, our findings indicate that GPT-series models perform more effectively on problems they have rephrased, suggesting a promising method for enhancing model capabilities.

Geometry problem solving (GPS) is a challenging mathematical reasoning task requiring multi-modal understanding, fusion, and reasoning. Existing neural solvers take GPS as a vision-language task but are short in the representation of geometry diagrams that carry rich and complex layout information. In this paper, we propose a layout-aware neural solver named LANS, integrated with two new modules: multimodal layout-aware pre-trained language module (MLA-PLM) and layout-aware fusion attention (LA-FA). MLA-PLM adopts structural-semantic pre-training (SSP) to implement global relationship modeling, and point-match pre-training (PMP) to achieve alignment between visual points and textual points. LA-FA employs a layout-aware attention mask to realize point-guided cross-modal fusion for further boosting layout awareness of LANS. Extensive experiments on datasets Geometry3K and PGPS9K validate the effectiveness of the layout-aware modules and superior problem-solving performance of our LANS solver, over existing symbolic and neural solvers. We have made our code and data publicly available.