Removed the 'device' arg from pipeline

Gemma.py

@@ -14,7 +14,8 @@ class GemmaLLM:
         model_id = "google/gemma-3-4b-it"
         model = AutoModelForCausalLM.from_pretrained(model_id, load_in_8bit=True, pad_token_id=0)
         tokenizer = AutoTokenizer.from_pretrained(model_id)
-        self.model = pipeline("text-generation", model=model, tokenizer=tokenizer, torch_dtype=torch.bfloat16, device="auto")
+        
+        self.model = pipeline("text-generation", model=model, tokenizer=tokenizer, torch_dtype=torch.bfloat16)
         
     @spaces.GPU
     def generate(self, message) -> str: