What Do Llamas Really Think? Revealing Preference Biases in Language Model Representations

Open in new window