Update model.py

model.py

@@ -11,11 +11,15 @@ if torch.cuda.is_available():
         model_id,
         torch_dtype=torch.float16,
         device_map='cuda',
+        
     ).to("cuda")
 else:
     model = AutoModelForCausalLM.from_pretrained(
         model_id,
-        torch_dtype=torch.float32,
+        torch_dtype=torch.float16,
+        load_in_4bit=True,
+        device_map='auto',
+        use_safetensors=False,
     )
 tokenizer = AutoTokenizer.from_pretrained(model_id)