The Eval Index / Red Teaming & Safety / #48
xinxuxin/keystone-bench
by xinxuxin · Red Teaming & Safety · updated 10d ago
A clinical decision-evidence benchmark: pull the one fact an answer rests on, and see whether the assistant's action moves with the evidence. Built on HealthBench.
72
momentum
374
stars
3
forks
#48
rank
ai-safetybenchmarkclinical-decision-supportdatasetevaluationhealthhealthbenchlarge-language-modelsllm-evaluationmedical-ai
View on GitHub →