1 code implementation • 25 May 2023 • Bruce W. Lee, Benedict Florance Arockiaraj, Helen Jin
We investigate the phenomenon of an LLM's untruthful response using a large set of 220 handcrafted linguistic features.