Update app.py

app.py

@@ -20,9 +20,10 @@ login(token=os.getenv("HuggingFaceFineGrainedReadToken"))
 
 model = AutoModelForCausalLM.from_pretrained(
     base_model_name,
-    load_in_4bit=True,  # ✅ Reduces memory
+    # load_in_4bit=True,  # ✅ Reduces memory, but requires a GPU
     torch_dtype=torch.float16,
-    device_map="auto"
+    # device_map="auto"  # Since I'm running on a free instance that doesn't have a GPU, I'll need to force CPU
+    device_map="cpu"
 )
 
 # Load LoRA adapter