fix streaming output

chat_app.py

@@ -40,17 +40,18 @@ def get_session_history(
     return store[(user_id, conversation_id)]
 
 prompt = ChatPromptTemplate.from_messages([
-    ("system", "You're an assistant who's good at everything"),
+    ("system", "[INST] You're an assistant who's good at everything"),
     MessagesPlaceholder(variable_name="history"),
-    ("human", "{question}"),
+    ("human", "{question} [/INST]"),
 ])
 
 model_id="mistralai/Mistral-7B-Instruct-v0.3"
 callbacks = [langchain_core.callbacks.StreamingStdOutCallbackHandler()]
 llm = HuggingFaceEndpoint(
     repo_id=model_id,
-    max_new_tokens=1024,
+    max_new_tokens=512,
     temperature=0.1,
+    repetition_penalty=1.03,
     callbacks=callbacks,
     streaming=True,
     huggingfacehub_api_token=os.getenv('HF_TOKEN'),
@@ -100,13 +101,16 @@ with gr.Blocks() as demo:
             {"ability": "everything", "question": question},
             config={"configurable": {"user_id": "123", "conversation_id": "1"}}
         )
-
         history.append({"role": "assistant", "content": ""})
-        for item in answer:
-            for character in item.content:
-                history[-1]['content'] += character
-                time.sleep(0.05)
-                yield history
+        for character in answer:
+            history[-1]['content'] += character
+            time.sleep(0.05)
+            yield history
+        # for item in answer:
+        #     for character in item.content:
+        #         history[-1]['content'] += character
+        #         time.sleep(0.05)
+        #         yield history
 
     submit_event = msg.submit(user, [msg, chatbot], [msg, chatbot], queue=True).then(
         bot, chatbot, chatbot