Wildtablebench

WildTableBench is a benchmark dataset for evaluating multimodal foundation models on table understanding in the wild. It contains 402 real-world table images collected from diverse domains and 928 questions across 5 categories and 17 subtypes. The dataset was created by author jzhuang and was last updated on Hugging Face in May 2026.

Use Cases

• Benchmarking model performance on numerical reasoning based on table images
• Evaluating cell locating capabilities in diverse real-world table formats
• Testing structural understanding of tables, such as identifying headers or relationships

Strengths

• 402 real-world table images provide a diverse testbed
• 928 questions across 5 categories and 17 subtypes offer structured evaluation

Limitations

• Dataset structure and column-level documentation are not fully described in the provided metadata
• Row count and file formats are unknown, which may limit suitability assessment

Provenance

Source

jzhuang via Hugging Face

Collection Method

402 real-world table images collected from diverse domains

Freshness

Last updated 2026-05-06 00:34:48