LangChain始终显示相同模板内容，重启内核无效的技术问题

使用LangChain的RetrievalQAWithSourcesChain时，无论是否尝试指定自定义模板，链始终显示相同的默认模板及示例内容，重启内核后问题仍存在。期望模板支持自定义或为空。

相关代码 #

chain = RetrievalQAWithSourcesChain.from_chain_type(
    llm,
    chain_type="stuff",
    retriever=db.as_retriever()
)
print(chain)

输出结果 #

Out:RetrievalQAWithSourcesChain(memory=None, callbacks=None, callback_manager=None, verbose=False, tags=None, metadata=None, combine_documents_chain=StuffDocumentsChain(memory=None, callbacks=None, callback_manager=None, verbose=False, tags=None, metadata=None, input_key='input_documents', output_key='output_text', llm_chain=LLMChain(memory=None, callbacks=None, callback_manager=None, verbose=False, tags=None, metadata=None, prompt=PromptTemplate(input_variables=['summaries', 'question'], output_parser=None, partial_variables={},
template='Given the following extracted parts of a long document and a question, create a final answer with references ("SOURCES"). \nIf you don\'t know the answer, just say that you don\'t know. Don\'t try to make up an answer.\nALWAYS return a "SOURCES" part in your answer.\n\nQUESTION: Which state/country\'s law governs the interpretation of the contract?\n=========\n
Content: This Agreement is governed by English law and the parties submit to the exclusive jurisdiction of the English courts in  relation to any dispute (contractual or non-contractual) concerning this Agreement save that either party may apply to any court for an  injunction or other relief to protect its Intellectual Property Rights.\nSource: 28-pl\nContent: No Waiver. Failure or delay in exercising any right or remedy under this Agreement shall not constitute a waiver of such (or any other)  right or remedy.\n\n11.7 Severability. The invalidity, illegality or unenforceability of any term (or part of a term) of this Agreement shall not affect the continuation  in force of the remainder of the term (if any) and this Agreement.\n\nWell I know this nation.\nSource: 34-pl\n=========\nFINAL ANSWER: The president did not mention Michael Jackson.\nSOURCES:\n\nQUESTION: {question}\n=========\n{summaries}\n=========\n
FINAL ANSWER:', template_format='f-string', validate_template=True),
llm=HuggingFaceHub(cache=None, verbose=False, callbacks=None, callback_manager=None, tags=None, metadata=None, client=InferenceAPI(api_url='https://api-inference.huggingface.co/pipeline/text2text-generation/mrm8488/t5-base-finetuned-question-generation-ap', task='text2text-generation', options={'wait_for_model': True, 'use_gpu': False}), repo_id='mrm8488/t5-base-finetuned-question-generation-ap', task=None, model_kwargs={'temperature': 0.9, 'max_new_tokens': 700}, huggingfacehub_api_token='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'), output_key='text', output_parser=StrOutputParser(), return_final_only=True, llm_kwargs={}), document_prompt=PromptTemplate(input_variables=['page_content', 'source'], output_parser=None, partial_variables={}, template='Content: {page_content}\nSource: {source}', template_format='f-string', validate_template=True), document_variable_name='summaries', document_separator='\n\n'), question_key='question', input_docs_key='docs', answer_key='answer', sources_answer_key='sources', return_source_documents=False, retriever=VectorStoreRetriever(tags=['Chroma', 'HuggingFaceEmbeddings'], metadata=None, vectorstore=<langchain.vectorstores.chroma.Chroma object at 0x7f56d774ae30>, search_type='similarity', search_kwargs={}), reduce_k_below_max_tokens=False, max_tokens_limit=3375)

解决方法 #

• 放弃使用from_chain_type的默认配置，手动构建链组件并传入自定义模板：

from langchain.chains import RetrievalQAWithSourcesChain, StuffDocumentsChain, LLMChain
from langchain.prompts import PromptTemplate

# 自定义主查询模板
custom_prompt = PromptTemplate(
    input_variables=["summaries", "question"],
    template="""根据以下文档片段和问题，生成带来源标注的最终答案。
若无法回答，请直接说明，禁止编造内容。必须包含SOURCES部分。

QUESTION: {question}
=========
{summaries}
=========
FINAL ANSWER: """
)

# 自定义文档片段格式化模板（可选）
document_prompt = PromptTemplate(
    input_variables=["page_content", "source"],
    template="内容: {page_content}\n来源: {source}"
)

# 构建LLM链
llm_chain = LLMChain(llm=llm, prompt=custom_prompt)

# 构建文档整合链
stuff_chain = StuffDocumentsChain(
    llm_chain=llm_chain,
    document_prompt=document_prompt,
    document_variable_name="summaries"
)

# 构建最终的检索QA链
chain = RetrievalQAWithSourcesChain(
    combine_documents_chain=stuff_chain,
    retriever=db.as_retriever()
)

print(chain)

• 上述代码中，可完全自定义主模板和文档格式化模板的内容、格式，替换默认的示例文本
• 需保证模板中的变量名（如summaries、question）与链组件中使用的变量名一致

内容的提问来源于stack exchange，提问作者Rishabh Gupta