This attack module uses a occupation dataset to generate a list of occupations related to demographic groups. This can be used to test the system with some level of stereotypical representation. To use this module effectively, add your API token to this endpoint llm-judge-openai-gpt4-annotator for evaluation.
To test this module, run this command and watch the magic happens:
This attack module uses a
occupation
dataset to generate a list of occupations related to demographic groups. This can be used to test the system with some level of stereotypical representation. To use this module effectively, add your API token to this endpointllm-judge-openai-gpt4-annotator
for evaluation.To test this module, run this command and watch the magic happens:
run_recipe -n 10 "test-module" "['bias-occupation']" "[<target>]"