Create app.py

app.py

@@ -0,0 +1,46 @@
+import streamlit as st
+from transformers import AutoModelForCausalLM, AutoTokenizer
+import torch
+
+# Load pre-trained DialoGPT-small model and tokenizer
+model_name = "microsoft/DialoGPT-small"
+model = AutoModelForCausalLM.from_pretrained(model_name)
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+
+# Set device to GPU if available for faster inference, otherwise fallback to CPU
+device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+model.to(device)
+
+# Initialize chat history
+if 'history' not in st.session_state:
+    st.session_state['history'] = []
+
+def generate_response(input_text):
+    # Encode the new user input, add end of string token
+    new_user_input_ids = tokenizer.encode(input_text + tokenizer.eos_token, return_tensors='pt').to(device)
+
+    # Append the new user input tokens to the chat history
+    bot_input_ids = torch.cat([torch.tensor(st.session_state['history']).to(device), new_user_input_ids], dim=-1) if st.session_state['history'] else new_user_input_ids
+
+    # Generate a response from the model
+    chat_history_ids = model.generate(bot_input_ids, max_length=1000, pad_token_id=tokenizer.eos_token_id, top_k=50, top_p=0.95, temperature=0.7)
+
+    # Decode the model's output and add it to the history
+    chat_history_ids = chat_history_ids[:, bot_input_ids.shape[-1]:]  # only take the latest generated tokens
+    bot_output = tokenizer.decode(chat_history_ids[0], skip_special_tokens=True)
+
+    # Update session state history with the new tokens
+    st.session_state['history'] = chat_history_ids[0].tolist()
+
+    return bot_output
+
+# Streamlit Interface
+st.title("Chat with DialoGPT")
+
+# Create input box for user
+user_input = st.text_input("You: ", "")
+
+if user_input:
+    # Generate and display the bot's response
+    response = generate_response(user_input)
+    st.write(f"Bot: {response}")