Int8WeightOnlyConfig

class torchao.quantization.Int8WeightOnlyConfig(group_size: Optional[int] = None, set_inductor_config: bool = True)

Configuration for applying int8 weight-only symmetric per-channel quantization to linear layers.