Confabulations detected by clustering semantically similar LLM responses.

Read more here: External Link