evaluation-dataset

HealthBench-Psych introduces a mental health subset of OpenAI's HealthBench benchmark to evaluate LLMs in mental health conversations, validated through clinician review and assessing 20 models as a reusable resource.