weiweiw
/
Langchain-Chatchat
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

update file format

This commit is contained in:
imClumsyPanda 2023-11-30 23:33:48 +08:00
parent f2d917d1e4
commit c34b4136c0
2 changed files with 44 additions and 31 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
configs/prompt_config.py.example
View File

@ -48,7 +48,7 @@ PROMPT_TEMPLATES = {
'<已知信息>{{ context }}</已知信息>\n' '<已知信息>{{ context }}</已知信息>\n'
'<问题>{{ question }}</问题>\n', '<问题>{{ question }}</问题>\n',
"Empty": # 搜不到知识库的时候使用 "empty": # 搜不到知识库的时候使用
'请你回答我的问题:\n' '请你回答我的问题:\n'
'{{ question }}\n\n', '{{ question }}\n\n',
}, },

31
server/chat/knowledge_base_chat.py
View File

@ -20,8 +20,14 @@ from server.knowledge_base.kb_doc_api import search_docs
async def knowledge_base_chat(query: str = Body(..., description="用户输入", examples=["你好"]), async def knowledge_base_chat(query: str = Body(..., description="用户输入", examples=["你好"]),
knowledge_base_name: str = Body(..., description="知识库名称", examples=["samples"]), knowledge_base_name: str = Body(..., description="知识库名称", examples=["samples"]),
top_k: int = Body(VECTOR_SEARCH_TOP_K, description="匹配向量数"), top_k: int = Body(VECTOR_SEARCH_TOP_K, description="匹配向量数"),
score_threshold: float = Body(SCORE_THRESHOLD, description="知识库匹配相关度阈值，取值范围在0-1之间，SCORE越小，相关度越高，取到1相当于不筛选，建议设置在0.5左右", ge=0, le=2), score_threshold: float = Body(
history: List[History] = Body([], SCORE_THRESHOLD,
description="知识库匹配相关度阈值，取值范围在0-1之间，SCORE越小，相关度越高，取到1相当于不筛选，建议设置在0.5左右",
ge=0,
le=2
),
history: List[History] = Body(
[],
description="历史对话", description="历史对话",
examples=[[ examples=[[
{"role": "user", {"role": "user",
@ -32,8 +38,14 @@ async def knowledge_base_chat(query: str = Body(..., description="用户输入",
stream: bool = Body(False, description="流式输出"), stream: bool = Body(False, description="流式输出"),
model_name: str = Body(LLM_MODELS[0], description="LLM 模型名称。"), model_name: str = Body(LLM_MODELS[0], description="LLM 模型名称。"),
temperature: float = Body(TEMPERATURE, description="LLM 采样温度", ge=0.0, le=1.0), temperature: float = Body(TEMPERATURE, description="LLM 采样温度", ge=0.0, le=1.0),
max_tokens: Optional[int] = Body(None, description="限制LLM生成Token数量，默认None代表模型最大值"), max_tokens: Optional[int] = Body(
prompt_name: str = Body("default", description="使用的prompt模板名称(在configs/prompt_config.py中配置)"), None,
description="限制LLM生成Token数量，默认None代表模型最大值"
),
prompt_name: str = Body(
"default",
description="使用的prompt模板名称(在configs/prompt_config.py中配置)"
),
request: Request = None, request: Request = None,
): ):
kb = KBServiceFactory.get_service_by_name(knowledge_base_name) kb = KBServiceFactory.get_service_by_name(knowledge_base_name)
@ -42,7 +54,8 @@ async def knowledge_base_chat(query: str = Body(..., description="用户输入",
history = [History.from_data(h) for h in history] history = [History.from_data(h) for h in history]
async def knowledge_base_chat_iterator(query: str, async def knowledge_base_chat_iterator(
query: str,
top_k: int, top_k: int,
history: Optional[List[History]], history: Optional[List[History]],
model_name: str = LLM_MODELS[0], model_name: str = LLM_MODELS[0],
@ -61,8 +74,8 @@ async def knowledge_base_chat(query: str = Body(..., description="用户输入",
) )
docs = search_docs(query, knowledge_base_name, top_k, score_threshold) docs = search_docs(query, knowledge_base_name, top_k, score_threshold)
context = "\n".join([doc.page_content for doc in docs]) context = "\n".join([doc.page_content for doc in docs])
if len(docs) == 0: ## 如果没有找到相关文档，使用Empty模板 if len(docs) == 0: # 如果没有找到相关文档，使用empty模板
prompt_template = get_prompt_template("knowledge_base_chat", "Empty") prompt_template = get_prompt_template("knowledge_base_chat", "empty")
else: else:
prompt_template = get_prompt_template("knowledge_base_chat", prompt_name) prompt_template = get_prompt_template("knowledge_base_chat", prompt_name)
input_msg = History(role="user", content=prompt_template).to_msg_template(False) input_msg = History(role="user", content=prompt_template).to_msg_template(False)
@ -80,14 +93,14 @@ async def knowledge_base_chat(query: str = Body(..., description="用户输入",
source_documents = [] source_documents = []
for inum, doc in enumerate(docs): for inum, doc in enumerate(docs):
filename = doc.metadata.get("source") filename = doc.metadata.get("source")
parameters = urlencode({"knowledge_base_name": knowledge_base_name, "file_name":filename}) parameters = urlencode({"knowledge_base_name": knowledge_base_name, "file_name": filename})
base_url = request.base_url base_url = request.base_url
url = f"{base_url}knowledge_base/download_doc?" + parameters url = f"{base_url}knowledge_base/download_doc?" + parameters
text = f"""出处 [{inum + 1}] [{filename}]({url}) \n\n{doc.page_content}\n\n""" text = f"""出处 [{inum + 1}] [{filename}]({url}) \n\n{doc.page_content}\n\n"""
source_documents.append(text) source_documents.append(text)
if len(source_documents) == 0: # 没有找到相关文档 if len(source_documents) == 0: # 没有找到相关文档
source_documents.append(f"""<span style='color:red'>未找到相关文档,该回答为大模型自身能力解答！</span>""") source_documents.append(f"<span style='color:red'>未找到相关文档,该回答为大模型自身能力解答！</span>")
if stream: if stream:
async for token in callback.aiter(): async for token in callback.aiter():