Add tests to perform the political compass test, following the work done in this article: From Pretraining Data to Language Models to Downstream Tasks: Tracking the Trails of Political Biases Leading to Unfair NLP Models, Github repo: here.
Basically, for LLMs, we have 62 statements to ask the LLM, and then the method can decide where in the political spectrum the LLM is (social values - liberal or conservative, and economic values - left or right aligned). Add a message saying that those 62 calls to the LLM are needed (and costs should be taken into consideration to run this test).
Add tests to perform the political compass test, following the work done in this article: From Pretraining Data to Language Models to Downstream Tasks: Tracking the Trails of Political Biases Leading to Unfair NLP Models, Github repo: here.
Basically, for LLMs, we have 62 statements to ask the LLM, and then the method can decide where in the political spectrum the LLM is (social values - liberal or conservative, and economic values - left or right aligned). Add a message saying that those 62 calls to the LLM are needed (and costs should be taken into consideration to run this test).