MarlinQQQTensor

class torchao.dtypes.MarlinQQQTensor(tensor_impl: AQTTensorImpl, block_size: Tuple[int, ...], shape: Size, quant_min: Optional[Union[int, float]] = None, quant_max: Optional[Union[int, float]] = None, zero_point_domain: ZeroPointDomain = ZeroPointDomain.INT, dtype=None, strides=None)

MarlinQQQ 量化张量子类，继承自 AffineQuantizedTensor 类。

要了解在 choose_qparams_and_quantize_affine_qqq、marllin qqq 量化的量化和反量化过程中发生的情况，请访问 https://github.com/pytorch/ao/blob/main/torchao/quantization/quant_primitives.py 并查看两个量化原始操作：choose_qparams_and_quantize_affine_qqq 和 dequantize_affine_qqq

dequantize() → Tensor

给定一个量化张量，对其进行反量化并返回反量化后的浮点张量。

classmethod from_hp_to_intx(input_float: Tensor, block_size: Tuple[int, ...], quant_min: Optional[int] = None, quant_max: Optional[int] = None, zero_point_domain: ZeroPointDomain = ZeroPointDomain.INT, _layout: Optional[Layout] = None)

将浮点张量转换为 Marlin QQQ 量化张量。