Update app.py

app.py

@@ -69,7 +69,6 @@ def load_model(model_name):
         # Only move to CUDA if it's not a quantized model
         if model_name not in quantized_models:
             model = model.to("cuda")
-            tokenizer = tokenizer.to("cuda")
         
         selected_model = model_name
     except Exception as e:
@@ -105,6 +104,9 @@ def interact(user_input, history, interaction_count):
         
         # Generate response using selected model
         input_ids = tokenizer(prompt, return_tensors='pt').input_ids
+        if model_name not in quantized_models:
+            input_ids.to("cuda")
+            
         chat_history_ids = model.generate(input_ids, max_new_tokens=100, pad_token_id=tokenizer.eos_token_id, temperature=0.1)
         response = tokenizer.decode(chat_history_ids[:, input_ids.shape[-1]:][0], skip_special_tokens=True)