Dr. Bench

Fits report-writing agents that need more than a single judge score. Borrow the reference bundle per task: query-specific rubrics, general report rubrics, trusted source links, and focus-anchor versus focus-deviation keywords.

What it asks the agent to do

• write a long report on an expert-curated question
• stay on topic as measured by focus keywords
• cite trustworthy sources from a reference list

What to borrow from it

grading designcase template

Caveats

No code repository seen, only a Hugging Face dataset; grading details come from the rubric fields rather than a published script; the name collides with DeepResearch Bench and DRBench.

Details

Links

• Paper