Search papers, labs, and topics across Lattice.
This paper introduces SynthDocBench, a synthetic benchmark designed for long-context visual document understanding, addressing the limitations of existing benchmarks by systematically controlling for factors like document length, layout complexity, modality, and question type. By employing a combinatorial design and generating documents through an LLM pipeline, the authors evaluate seven state-of-the-art vision language models (VLMs) and identify critical failure modes that existing benchmarks fail to reveal, including significant performance drops with increased document length and a systematic positional sensitivity. The findings indicate that current VLMs may be overfitting to artifacts in existing benchmarks, highlighting the need for more robust evaluation methods in this domain.
Current VLMs show alarming performance drops in long-context scenarios, revealing that they may be overfitting to benchmark artifacts rather than demonstrating true understanding.
Vision language models (VLMs) have achieved strong performance on visual document understanding benchmarks such as DocVQA, ChartQA, and MMLongBench-Doc. However, real-world documents combine multiple factors such as length, layout complexity, modality, and question difficulty, which makes it difficult to attribute model failures to specific causes. We introduce SynthDocBench, a fully synthetic benchmark for long-context visual document understanding that systematically controls factors including document length, layout structure, modality composition, and question type. The benchmark is constructed using a combinatorial design, each factor is varied independently across generated documents, enabling controlled analysis of model behavior. Documents are generated end to end using an LLM pipeline across six layout archetypes, with a 40 percent random override to prevent models from exploiting spurious correlations. Additionally, SynthDocBench spans long-context documents with substantially greater length and structural diversity than existing benchmarks. Evaluating seven frontier VLMs, we uncover three failure modes that existing benchmarks cannot surface: sharp degradation with document length, a systematic positional sensitivity in which the middle third of a document is hardest for five of six models and five of six models show a negative Early-to-Late trend (steepest decline: 8.3 percentage points), and breakdown of chart comprehension in long-document settings. These results suggest that current models may be overfitting to benchmark artifacts rather than achieving robust long-context visual document understanding.
