WildDoc: A Benchmark for Real-World Document Understanding by Vision-Language Models

WildDoc is a dataset created by ByteDance to evaluate the document understanding capabilities of vision-language models in real-world scenarios. It is designed to facilitate the understanding of documents in the wild, as described on its project homepage. The dataset was last updated on May 19, 2025.

Use Cases

• Benchmarking model performance on real-world document understanding tasks based on the dataset's stated purpose.
• Training models to interpret unstructured documents based on the 'documents in the wild' concept.
• Evaluating the robustness of vision-language models against diverse document formats implied by the 'wild' context.

Strengths

• Created by ByteDance, a major technology organization.
• Specifically designed for evaluating real-world document understanding capabilities.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Last updated 2025-05-19 03:13:44; freshness should be verified.

Provenance

Source

ByteDance

Freshness

Last updated 2025-05-19 03:13:44.