1  ---
2  features:
3    - |
4      Add a new FaithfulnessEvaluator component that can be used to evaluate faithfulness / groundedness / hallucinations of LLMs in a RAG pipeline.
5      Given a question, a list of retrieved document contents (contexts), and a predicted answer, FaithfulnessEvaluator returns a score ranging from 0 (poor faithfulness) to 1 (perfect faithfulness).
6      The score is the proportion of statements in the predicted answer that could by inferred from the documents.