israel-adewuyi/med-safety-bench-reproduced
Med Safety Bench Dataset This dataset is a collection of harmful medical requests and safe responses, originally sourced from the AI4LIFE-GROUP/med-safety-bench GitHub repository. The dataset is intended for research purposes related to the safety of medical AI models. Data Structure The dataset contains the following columns: harmful_medical_request: The harmful medical request. safe_response: The safe response to the request. source: Indicates whether the data… See the full description on the dataset page: https://huggingface.co/datasets/israel-adewuyi/med-safety-bench-reproduced.
Med Safety Bench Dataset
This dataset is a collection of harmful medical requests and safe responses, originally sourced from the AI4LIFE-GROUP/med-safety-bench GitHub repository.
The dataset is intended for research purposes related to the safety of medical AI models.
Data Structure
The dataset contains the following columns:
harmful_medical_request: The harmful medical request.safe_response: The safe response to the request.source: Indicates whether the data originated from GPT-4 or Llama 2.
Original Source
The original data and more information can be found at the Med Safety Bench GitHub repository.