Test Benchmark 2

This most simple benchmark is used to test the benchmarking system. It contains a single page containing Martin Luther's Ninety-Five Theses. The task is to extract the theses from the text. The evaluation is based on the number of correctly extracted theses.

Benchmark Setup

• The page is a historical document.
• The task is to extract the theses from the text.
• The model should extract the theses and save them in a given JSON structure.
• The evaluation is based on the number of correctly extracted theses.

Test Results