Improving Preference Extraction In LLMs By Identifying Latent Knowledge Through Classifying Probes

Re-assign community