Update app.py

app.py

@@ -42,6 +42,7 @@ from langchain.vectorstores.utils import filter_complex_metadata
 import fitz
 from PIL import Image
 from langchain.vectorstores import FAISS
+import transformers
 
 user_session_id = uuid.uuid4()
 
@@ -136,6 +137,7 @@ def load_model(_docs):
     template = generate_prompt("""{context}  Question: {question} """,system_prompt=SYSTEM_PROMPT,) #Enter memory here!
 
     prompt = PromptTemplate(template=template, input_variables=["context",  "question"]) #Add history here
+    streamer = transformers.TextIteratorStreamer(tokenizer, timeout=10., skip_prompt=True, skip_special_tokens=True)
 
     qa_chain = RetrievalQA.from_chain_type(
         llm=llm,
@@ -144,6 +146,7 @@ def load_model(_docs):
         return_source_documents=True,
         chain_type_kwargs={"prompt": prompt,
                            "verbose": False,
+                           streamer=streamer,
                            #"memory": ConversationBufferMemory(
                               #memory_key="history",
                               #input_key="question",