Update app.py

app.py

@@ -53,8 +53,7 @@ with st.container():
         with st.chat_message(message["role"]):
             st.write(message["content"])
             
-    tokenized_chat = tokenizer.apply_chat_template(st.session_state.chat_history, tokenize=True, add_generation_prompt=True, return_tensors="pt")
-    st.write(tokenizer.decode(tokenized_chat[0]))
+    
     #Set up input text field
     input_text = st.chat_input(placeholder="Here you can chat with our hotel booking model.")
 
@@ -65,8 +64,10 @@ with st.container():
 
         #chat_response = demo_chat.demo_chain(input_text=input_text, memory=st.session_state.memory, model= chat_model)
         #first_answer = chat_response.split("Human")[0] #Because of Predict it prints the whole conversation.Here we seperate the first answer only.
+        tokenized_chat = tokenizer.apply_chat_template(st.session_state.chat_history, tokenize=True, add_generation_prompt=True, return_tensors="pt")
+        st.write(tokenizer.decode(tokenized_chat[0]))
         outputs = model.generate(tokenized_chat, max_new_tokens=128) 
-        first_answer = tokenizer.decode(outputs[0])
+        first_answer = tokenizer.decode(outputs[0],skip_special_tokens=True)
         
         with st.chat_message("assistant"):
             st.write(first_answer)