Update app.py

app.py

@@ -1,23 +1,33 @@
 import streamlit as st
-from transformers import pipeline
+from transformers import AutoModelForCausalLM, AutoTokenizer
+import torch
 
-# Load the conversational pipeline
-conversational_pipeline = pipeline("conversational")
+# Load DialoGPT model and tokenizer
+tokenizer = AutoTokenizer.from_pretrained("microsoft/DialoGPT-medium")
+model = AutoModelForCausalLM.from_pretrained("microsoft/DialoGPT-medium")
 
 # Streamlit app header
 st.set_page_config(page_title="Conversational Model Demo", page_icon="🤖")
 st.header("Conversational Model Demo")
 
+# Initialize chat history
+chat_history_ids = None
+
 # Input for user message
 user_message = st.text_input("You:", "")
 
 if st.button("Send"):
-    # Format the conversation for the conversational pipeline
-    conversation_history = [{"role": "system", "content": "You are an AI assistant."},
-                           {"role": "user", "content": user_message}]
+    # Encode the new user input, add the eos_token and return a tensor in PyTorch
+    new_user_input_ids = tokenizer.encode(user_message + tokenizer.eos_token, return_tensors='pt')
+
+    # Append the new user input tokens to the chat history
+    bot_input_ids = torch.cat([chat_history_ids, new_user_input_ids], dim=-1) if chat_history_ids is not None else new_user_input_ids
+
+    # Generate a response while limiting the total chat history to 1000 tokens
+    chat_history_ids = model.generate(bot_input_ids, max_length=1000, pad_token_id=tokenizer.eos_token_id)
 
-    # Get the model's response
-    model_response = conversational_pipeline(conversation_history)[0]['generated_text']
+    # Pretty print last output tokens from the bot
+    model_response = tokenizer.decode(chat_history_ids[:, bot_input_ids.shape[-1]:][0], skip_special_tokens=True)
 
     # Display the model's response
     st.text_area("Model:", model_response, height=100)