PyPI Stats

Search

All packages
Top packages

Track packages

keystone-bench


PyPI page
Home page
Author: Xin Xu
License: MIT
Summary: Paired evidence-perturbation stress tests for clinical chat assistants and their rubric graders (built on HealthBench)
Latest version: 0.4.0
Required dependencies: httpx
Optional dependencies: inspect-ai | numpy | pytest | scipy | statsmodels

Downloads last day: 0
Downloads last week: 4
Downloads last month: 155
7-day smoothing