Update functions.py

functions.py

@@ -31,7 +31,7 @@ from langchain.text_splitter import RecursiveCharacterTextSplitter
 from langchain.chat_models import ChatOpenAI
 from langchain.callbacks.base import CallbackManager
 from langchain.callbacks.streaming_stdout import StreamingStdOutCallbackHandler
-from langchain.chains import ConversationalRetrievalChain, QAGenerationChain
+from langchain.chains import ConversationalRetrievalChain, QAGenerationChain, RetrievalQA
 from langchain.memory import ConversationBufferMemory
 
 from langchain.prompts.chat import (
@@ -224,19 +224,23 @@ def embed_text(query,embedding_model,_docsearch):
     chat_history = []
     # llm = OpenAI(temperature=0)
     chat_llm = ChatOpenAI(streaming=True,
-                          model_name = 'gpt-3.5-turbo',
+                          model_name = 'gpt-4',
                           callback_manager=CallbackManager([StreamingStdOutCallbackHandler()]), 
                           verbose=True, 
                           temperature=0
                          )
 
-    chain = ConversationalRetrievalChain.from_llm(chat_llm, 
-                                                  retriever= _docsearch.as_retriever(),
-                                                  # condense_question_prompt = load_prompt(),
-                                                  memory = memory,
-                                                  return_source_documents=True)
+    chain = RetrievalQA.from_chain_type(llm=chat_llm, chain_type="stuff", 
+                                 retriever=_docsearch.as_retriever(), 
+                                 return_source_documents=True)
     
-    answer = chain({"question": query})
+    # chain = ConversationalRetrievalChain.from_llm(chat_llm, 
+    #                                               retriever= _docsearch.as_retriever(),
+    #                                               # condense_question_prompt = load_prompt(),
+    #                                               memory = memory,
+    #                                               return_source_documents=True)
+    
+    answer = chain({"query": query})
 
     return answer