Awesome Table Understanding: Curated Directory of Research Benchmarks

Aggregating multiple benchmarks for table understanding, this repository by esborisova was updated in September 2025. It categorizes resources into tasks such as table structure recognition, table-to-text, and table question answering.

Use Cases

• Benchmarking table structure recognition models using linked ground truth annotations
• Training table-to-text systems using referenced natural language descriptions
• Evaluating table question answering performance against provided query-response pairs

Strengths

• Updated as of September 2025
• Categorized by specific table understanding tasks
• Released under MIT license

Limitations

• Does not contain raw data files directly
• Relies on external links which may become stale
• No unified data format across the listed resources

Provenance

Source

esborisova (GitHub)

Collection Method

Curated list

Freshness

Last updated September 2025