Update app.py

app.py

@@ -3,14 +3,13 @@ from transformers import AutoModelForCausalLM, AutoTokenizer
 import torch, re, json
 from datetime import datetime
 
-device = "cuda"  # the device to load the model onto
 
 # Load model and tokenizer
 model = AutoModelForCausalLM.from_pretrained(
     "Qwen/Qwen1.5-0.5B-Chat",
     torch_dtype="auto",
     device_map="auto",
-).to(device)
+)
 tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen1.5-0.5B-Chat")
 
 if 'recipe' not in st.session_state:
@@ -120,7 +119,7 @@ def generate_recipe(user_inputs):
         )
 
         # Tokenize and move to the correct device
-        model_inputs = tokenizer([text], return_tensors="pt").to(device)
+        model_inputs = tokenizer([text], return_tensors="pt")
         torch.cuda.empty_cache()
         with torch.no_grad():
             generated_ids = model.generate(