Remove unused pipeline

prompting/llms/hf_llm.py

@@ -26,7 +26,6 @@ def __init__(
         self.valid_generation_params = set(
             AutoModelForCausalLM.from_pretrained(model_id).generation_config.to_dict().keys()
         )
-        self.llm = pipeline("text-generation", model=self.model, tokenizer=self.tokenizer)
 
     @torch.inference_mode()
     def generate(