LLAMADRS: Evaluating Open-Source LLMs on Real Clinical Interviews--To Reason or Not to Reason?

Gaoussou Youssouf Kebe, Jeffrey M. Girard, Einat Liebenthal +3

Large language models (LLMs) excel on many NLP benchmarks, but their behavior on real-world, semi-structured prediction remains underexplored. We present LlaMADRS, a benchmark for…