Update app.py

app.py

@@ -1,24 +1,33 @@
 import streamlit as st
 from transformers import AutoModelForCausalLM, AutoTokenizer
+from accelerate import Accelerator  # Import Accelerator from the accelerate library
 
 def main():
     st.title("Chatbot with Hugging Face Model")
 
     model_id = "mistralai/Mixtral-8x7B-Instruct-v0.1"
     tokenizer = AutoTokenizer.from_pretrained(model_id)
-    model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto")
+    
+    # Create an Accelerator instance
+    accelerator = Accelerator()
+
+    # Use the Accelerator for initializing the model
+    model = AutoModelForCausalLM.from_pretrained(model_id, device_map=accelerator.device)
 
     user_input = st.text_input("User Input:", "What is your favourite condiment?")
 
     if st.button("Generate Response"):
         messages = [
             {"role": "user", "content": user_input},
-            {"role": "assistant", "content": "Placeholder assistant message"}  # You can modify this as needed
+            {"role": "assistant", "content": "Placeholder assistant message"}
         ]
 
-        inputs = tokenizer.apply_chat_template(messages, return_tensors="pt").to("cuda")
+        inputs = tokenizer.apply_chat_template(messages, return_tensors="pt").to(accelerator.device)
+
+        # Use the Accelerator for generating outputs
+        with accelerator.device():
+            outputs = model.generate(inputs, max_new_tokens=20)
 
-        outputs = model.generate(inputs, max_new_tokens=20)
         response = tokenizer.decode(outputs[0], skip_special_tokens=True)
 
         st.text_area("Assistant's Response:", response)