Skip to main content

​
Functions

​
FUNC check_answerability 

check_answerability(question: str, documents: collections.abc.Iterable[Document], context: ChatContext, backend: AdapterMixin) -> float
Test a user’s question for answerability. Intrinsic function that checks whether the question in the last user turn of a chat can be answered by a provided set of RAG documents. :param context: Chat context containing the conversation thus far :param question: Question that the user has posed in response to the last turn in context. :param documents: Document snippets retrieved that may or may not answer the indicated question. :param backend: Backend instance that supports adding the LoRA or aLoRA adapters for answerability checks :return: Answerability score as a floating-point value from 0 to 1.

​
FUNC rewrite_question 

rewrite_question(question: str, context: ChatContext, backend: AdapterMixin) -> float
Rewrite a user’s question for retrieval. Intrinsic function that rewrites the question in the next user turn into a self-contained query that can be passed to the retriever. :param context: Chat context containing the conversation thus far :param question: Question that the user has posed in response to the last turn in context. :param backend: Backend instance that supports adding the LoRA or aLoRA adapters :return: Rewritten version of question.

​
FUNC find_citations 

find_citations(response: str, documents: collections.abc.Iterable[Document], context: ChatContext, backend: AdapterMixin) -> list[dict]
Find information in documents that supports an assistant response. Intrinsic function that finds sentences in RAG documents that support sentences in a potential assistant response to a user question. :param context: Context of the dialog between user and assistant at the point where the user has just asked a question that will be answered with RAG documents :param response: Potential assistant response :param documents: Documents at were used to generate response. These documents should set the doc_id field; otherwise the intrinsic will be unable to specify which document was the source of a given citation. :param backend: Backend that supports one of the adapters that implements this intrinsic. :return: List of records with the following fields:
  • response_begin
  • response_end
  • response_text
  • citation_doc_id
  • citation_begin
  • citation_end
  • citation_text Begin and end offsets are character offsets into their respective UTF-8 strings.

​
FUNC check_context_relevance 

check_context_relevance(question: str, document: Document, context: ChatContext, backend: AdapterMixin) -> float
Test whether a document is relevant to a user’s question. Intrinsic function that checks whether a single document contains part or all of the answer to a user’s question. Does not consider the context in which the question was asked. :param context: The chat up to the point where the user asked a question. :param question: Question that the user has posed. :param document: A retrieved document snippet :param backend: Backend instance that supports the adapters that implement this intrinsic :return: Context relevance score as a floating-point value from 0 to 1.

​
FUNC flag_hallucinated_content 

flag_hallucinated_content(response: str, documents: collections.abc.Iterable[Document], context: ChatContext, backend: AdapterMixin) -> float
Flag potentially-hallucinated sentences in an agent’s response. Intrinsic function that checks whether the sentences in an agent’s response to a user question are faithful to the retrieved document snippets. Sentences that do not align with the retrieved snippets are flagged as potential hallucinations. :param context: A chat log that ends with a user asking a question :param response: The assistant’s response to the user’s question in the last turn of context :param documents: Document snippets that were used to generate response :param backend: Backend instance that supports the adapters that implement this intrinsic :return: List of records with the following fields:
  • response_begin
  • response_end
  • response_text
  • faithfulness_likelihood
  • explanation

​
FUNC rewrite_answer_for_relevance 

rewrite_answer_for_relevance(rewrite_threshold: float = 0.5) -> str
Rewrite an assistant answer to improve relevance to the user’s question. :param context: A chat log that ends with a user asking a question :param response: The assistant’s response to the user’s question in the last turn of context :param documents: Document snippets that were used to generate response :param backend: Backend instance that supports the adapters that implement this intrinsic :param rewrite_threshold: Number between 0.0 and 1.0 that determines how eagerly to skip rewriting the assistant’s answer for relevance. 0.0 means never rewrite and 1.0 means always rewrite. :returns: Either the original response, or a rewritten version of the original response.