from transformers import PretrainedConfig

class BitNetConfig(PretrainedConfig):
    model_type = "bitnet"
    def __init__(self, bit_width=1.58, group_size=64, **kwargs):
        super().__init__(**kwargs)
        self.bit_width = bit_width    # 权重量化位数
        self.group_size = group_size  # 量化分组大小