Function quantize

tensorrt_llm/quantization/functional.py:755–766

(input: Tensor,
             scale_factor: Tensor,
             dtype: str,
             axis: int = -1)

def quantize(input: Tensor,
             scale_factor: Tensor,
             dtype: str,
             axis: int = -1) -> Tensor:
    layer = default_trtnet().add_quantize(input.trt_tensor,
                                          scale_factor.trt_tensor,
                                          str_dtype_to_trt(dtype))
    layer.axis = axis

    output = _create_tensor(layer.get_output(0), layer)

    return output


def dequantize(input: Tensor,