ModelScope中，我在推理的时候我每次只输入了一个问题，代码会自动再输一次，是为什么？

ModelScope中，我在推理的时候是人工输入问题，我每次只输入了一个问题，但是代码会自动再输入一个问题
其实我只输入了，我感冒了是internlm-20b-chat这个模型
以下是我的推理启动脚本：class InferArguments:
model_type: str = field(
default='internlm-20b-chat',
metadata={'choices': list(MODEL_MAPPING.keys())})
sft_type: str = field(
default='lora', metadata={'choices': ['lora', 'full']})
template_type: str = field(
default= 'internlm', metadata={'choices': list(TEMPLATE_MAPPING.keys())})
ckpt_dir: str = 'output/internlm-20b-chat/v0-20231024-170814/checkpoint-140'
eval_human: bool = True # False: eval val_dataset

seed: int = 42
dtype: str = field(
    default='bf16', metadata={'choices': ['bf16', 'fp16', 'fp32']})
ignore_args_error: bool = False  # True: notebook compatibility

dataset: str = field(
    default='QA-gen',
    metadata={'help': f'dataset choices: {list(DATASET_MAPPING.keys())}'})
dataset_split_seed: int = 42
dataset_test_ratio: float = 0.2
show_dataset_sample: int = 10
system: str = 'you are a helpful assistant!'
max_length: Optional[int] = 2048

quantization_bit: int = field(default=0, metadata={'choices': [0, 4, 8]})
bnb_4bit_comp_dtype: str = field(
    default=None, metadata={'choices': ['fp16', 'bf16', 'fp32']})
bnb_4bit_quant_type: str = field(
    default='nf4', metadata={'choices': ['fp4', 'nf4']})
bnb_4bit_use_double_quant: bool = True

max_new_tokens: int = 1024
do_sample: bool = True
temperature: float = 0.9
top_k: int = 20
top_p: float = 0.9
repetition_penalty: float = 1.

# other
use_flash_attn: Optional[bool] = field(
    default=None,
    metadata={
        'help':
        "This parameter is used only when model_type.startswith('qwen')"
    })
use_streamer: bool = True
merge_lora_and_save: bool = True
save_generation_config: bool = True

def init_argument(self):
    # Can be manually initialized, unlike __post_init__
    if self.template_type is None:
        self.template_type = MODEL_MAPPING[self.model_type].get(
            'template', 'default')
        logger.info(f'Setting template_type: {self.template_type}')

    self.torch_dtype, _, _ = select_dtype(self)
    if self.bnb_4bit_comp_dtype is None:
        self.bnb_4bit_comp_dtype = self.dtype
    self.bnb_4bit_compute_dtype, self.load_in_4bit, self.load_in_8bit = select_bnb(
        self)

    if self.use_flash_attn is None:
        self.use_flash_attn = 'auto'