[RAG] 全部LLM硬编码从deepseek-r1改为deepseek-v3；修复知识库问答source格式不匹配导致检索结果被丢弃

langchain-chat/server/chat/chat.py

@@ -62,7 +62,7 @@ async def chat(query: str = Body(..., description="用户输入", examples=["恼
 
         if prompt_name == "Search Summary":
             model = get_ChatOpenAI(
-                model_name=LLM_MODELS[1],
+                model_name=LLM_MODELS[0],
                 temperature=temperature,
                 max_tokens=max_tokens,
                 callbacks=callbacks,

langchain-chat/server/chat/chat_comparison.py

@@ -46,7 +46,7 @@ async def chat_comparison_test(
                 executor.submit(
                     get_llm_model_response,
                     strategy_name="query rewrite",
-                    llm_model_name=LLM_MODELS[1],
+                    llm_model_name=LLM_MODELS[0],
                     template_prompt_name="extract_key_points",
                     prompt_param_dict={"time": datetime.now().strftime("%Y%m%d"), "context": context, "content": content},
                     temperature=0.01,

langchain-chat/server/chat/gen_abstract.py

@@ -44,7 +44,7 @@ async def gen_abstract(
     try:
         article_abstract = get_llm_model_response(
             strategy_name="gen_abstract",
-            llm_model_name=LLM_MODELS[1],
+            llm_model_name=LLM_MODELS[0],
             template_prompt_name="gen_abstract",
             prompt_param_dict={
                 "context": context_summary,  # 使用摘要或原文

langchain-chat/server/chat/gen_conclusion.py

@@ -44,7 +44,7 @@ async def gen_conclusion(
     try:
         article_conclusion = get_llm_model_response(
             strategy_name="gen_conclusion",
-            llm_model_name=LLM_MODELS[1],
+            llm_model_name=LLM_MODELS[0],
             template_prompt_name="gen_conclusion",
             prompt_param_dict={
                 "context": context_summary,  # 使用摘要或原文

langchain-chat/server/chat/gen_keywords.py

@@ -43,7 +43,7 @@ async def gen_keywords(
     try:
         article_keywords = get_llm_model_response(
             strategy_name="gen_keywords",
-            llm_model_name=LLM_MODELS[1],
+            llm_model_name=LLM_MODELS[0],
             template_prompt_name="gen_keywords",
             prompt_param_dict={
                 "context": context_summary,  # 使用摘要或原文

langchain-chat/server/chat/gen_paragraph.py

@@ -44,7 +44,7 @@ async def gen_paragraph(
     try:
         article_paragraph = get_llm_model_response(
             strategy_name="gen_paragraph",
-            llm_model_name=LLM_MODELS[1],
+            llm_model_name=LLM_MODELS[0],
             template_prompt_name="gen_paragraph",
             prompt_param_dict={
                 "context": context_summary,  # 使用摘要或原文

langchain-chat/server/chat/gen_title.py

@@ -19,7 +19,7 @@ async def gen_title(
                                             "content": "我们来玩成语接龙，我先来，生龙活虎"},
                                             {"role": "assistant", "content": "虎头虎脑"}]]
                                         ),
-            model_name: Optional[str] = Body(LLM_MODELS[1], description="LLM 模型名称。"),
+            model_name: Optional[str] = Body(LLM_MODELS[0], description="LLM 模型名称。"),
 ):
     """
     根据一轮对话历史生成简洁标题\n
@@ -32,7 +32,7 @@ async def gen_title(
     if model_name == "R1-70B":
         model_name = DEEPSEEK_MODELS[1]
     elif model_name == "QIANWEN":
-        model_name = LLM_MODELS[1]
+        model_name = LLM_MODELS[0]
     else:
         model_name = model_name
 

langchain-chat/server/chat/knowledge_base_chat.py

@@ -138,7 +138,7 @@ async def knowledge_base_chat(query: str = Body(..., description="用户输入",
         if isinstance(max_tokens, int) and max_tokens <= 0:
             max_tokens = None
         if prompt_name == "policy_chat":
-            model_name = LLM_MODELS[1]
+            model_name = LLM_MODELS[0]
         model = get_ChatOpenAI(
             model_name=model_name,
             temperature=temperature,

langchain-chat/server/chat/search_engine_chat.py

@@ -364,7 +364,7 @@ async def search_engine_chat(uid: Optional[str]=Body(None, description="userID")
                                                                 "content": "虎头虎脑"}]]
                                                            ),
                              stream: bool = Body(False, description="流式输出"),
-                             model_name: str = Body(LLM_MODELS[1], 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),
                              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"),
                              prompt_name: str = Body("default",
@@ -483,7 +483,7 @@ async def search_engine_chat(uid: Optional[str]=Body(None, description="userID")
                                           search_engine_name: str,
                                           top_k: int,
                                           history: Optional[List[History]],
-                                          model_name: str = LLM_MODELS[1],
+                                          model_name: str = LLM_MODELS[0],
                                           prompt_name: str = prompt_name,
                                           kgo_search_type: str = kgo_search_type
                                           ) -> AsyncIterable[str]:

langchain-chat/server/chat/self_kb_chat.py

@@ -72,7 +72,7 @@ async def self_kb_chat(
     async def knowledge_base_chat_iterator(
             query: str,
             model_name: str = LLM_MODELS[0],
-            model_name1: str = LLM_MODELS[1],
+            model_name1: str = LLM_MODELS[0],
             prompt_name: str = "self_default",
     ) -> AsyncIterable[str]:
         nonlocal fileNames, history