Update app.py

app.py

@@ -6,11 +6,23 @@ from langchain.chains.question_answering import load_qa_chain
 from langchain.llms import OpenAI
 from gradio import gradio as gr
 from langchain.chat_models import ChatOpenAI
-#from langchain.memory import ConversationBufferMemor
-from langchain.schema import AIMessage, HumanMessage
+
 from langchain import PromptTemplate, LLMChain
 from langchain.llms import TextGen
 from langchain.cache import InMemoryCache
+
+from langchain.prompts.chat import (
+    ChatPromptTemplate,
+    SystemMessagePromptTemplate,
+    AIMessagePromptTemplate,
+    HumanMessagePromptTemplate,
+)
+from langchain.schema import (
+    AIMessage,
+    HumanMessage,
+    SystemMessage
+)
+
 import time
 import langchain
 import os
@@ -23,19 +35,24 @@ embeddings = HuggingFaceEmbeddings(model_name="BAAI/bge-large-en")
 # 加载数据
 #docsearch = FAISS.from_texts(texts, embeddings)
 docsearch = FAISS.load_local("./faiss_index", embeddings)
+chain = load_qa_chain(OpenAI(temperature=0,model_name="gpt-3.5-turbo", verbose=True), chain_type="stuff",verbose=True)
 
 
-chain = load_qa_chain(OpenAI(temperature=0,model_name="gpt-3.5-turbo", verbose=True), chain_type="stuff",verbose=True)
 
-prompt = "您是回答所有ANSYS软件使用查询的得力助手,如果所问的内容不在范围内，请回答“您提的问题不在本知识库内，请重新提问”，所有问题必需用中文回答"
+#template="您是回答所有ANSYS软件使用查询的得力助手,如果所问的内容不在范围内，请回答您提的问题不在本知识库内，请重新提问. {input_language} to {output_language}."
+#system_message_prompt = SystemMessagePromptTemplate.from_template(template)
+#human_template="{text}"
+#human_message_prompt = HumanMessagePromptTemplate.from_template(human_template
+
+prompt = "您是回答所有ANSYS软件使用查询的得力助手,如果所问的内容不在范围内，请回答您提的问题不在本知识库内，请重新提问，所有问题必需用中文回答"
 
 def predict(message, history):
-    history_openai_format = []
-    for human, assistant in history:
-        history_openai_format.append({"role": "system", "content": prompt })
-        history_openai_format.append({"role": "user", "content": human })
-        history_openai_format.append({"role": "assistant", "content":assistant})
-    history_openai_format.append({"role": "user", "content": message})
+    history_langchain_format = []
+    for human, ai in history:
+        history_langchain_format.append(SystemMessage(content=prompt))
+        history_langchain_format.append(HumanMessage(content=human))
+        history_langchain_format.append(AIMessage(content=ai))
+    history_langchain_format.append(HumanMessage(content=message))
     docs = docsearch.similarity_search(message)
     response = chain.run(input_documents=docs, question=message)