Med-PaLM 2 evaluates Large Language Models using a framework for long-form question answering that incorporates nine physician-validated evaluation axes and adversarial datasets to test model safety and accuracy.

Authors

• Person: Not available Organization: arXiv
  A Comprehensive Benchmark and Evaluation Framework for Multi ...

Sources

• A Comprehensive Benchmark and Evaluation Framework for Multi ... arxiv.org arXiv via serper

Referenced by nodes (1)

• Large Language Models concept