LiteRT 8 位量化规范

以下文档概述了 LiteRT 8 位架构的规范量化方案这有助于硬件开发者提供硬件支持，以便通过量化 LiteRT 模型进行推理。

规范摘要

我们仅提供了具体规范，只能对行为。我们也知道不同的硬件可能会造成轻微偏差导致实现不精确的规范。然而，在大多数情况下，这样做是可以接受的（我们会提供一套据我们所知，测试中包含我们针对每项操作机器学习的本质（以及深度学习导致无法提供任何硬性保证。

8 位量化使用以下函数近似计算浮点值公式。

\[real\_value = (int8\_value - zero\_point) \times scale\]

每个轴（在转换操作中也称为每个渠道）或每个张量权重由 int8：在 [-127, 127] 范围内具有零点相等的二元补数值设置为 0。每个张量的激活/输入由 int8 二补数表示值在 [-128, 127] 范围内，零点在 [-128, 127] 范围内。

特定操作还存在其他例外情况，如下所述。

有符号整数与无符号整数

LiteRT 量化将主要优先考虑用于以下用途的工具和内核： int8 量化（针对 8 位）。这是为了方便对称量化由等于 0 的零点表示。此外，许多后端针对累积 int8xint8 进行了其他优化。

每个轴与每个张量

每张量量化意味着，每个张量都有一个量级和/或零点整个张量。每轴量化意味着将有一个刻度和/或 quantized_dimension 中每个切片 zero_point。量化维度指定缩放和零点数的张量形状的维度对应的版本。例如，一个张量 t，dims=[4, 3, 2, 1] 和量化参数：scale=[1.0, 2.0, 3.0]、zero_point=[1, 2, 3]、 quantization_dimension=1 将在 t 的第二个维度中进行量化：

t[:, 0, :, :] will have scale[0]=1.0, zero_point[0]=1
t[:, 1, :, :] will have scale[1]=2.0, zero_point[1]=2
t[:, 2, :, :] will have scale[2]=3.0, zero_point[2]=3

通常，quantized_dimension 为以下项的权重的 output_channel：但从理论上讲，它可以是每个内核实现中的 dot-product，可实现更高的量化粒度且不会影响性能这在很大程度上提高了准确性。

TFLite 支持每轴的操作数量不断增加。截至本文档均支持 Conv2d 和 DepthwiseConv2d。

对称与不对称

激活是不对称的：激活函数可以在带符号的 int8 范围 [-128, 127]。许多激活在本质上是不对称的，零点是获得额外二进制精度。由于激活量仅乘以常数，则常量零点值可以得到大幅优化。

权重是对称的：强制使零点等于 0。权重值为乘以动态输入值和激活值。这意味着将权重零点乘以激活值。通过强制将零点设为 0，我们可以避免这一开销。

数学说明：这类似于 arXiv:1712.05877（差异部分除外）我们允许缩放值按轴显示。这很容易概括，因为如下：

$A$ 是量化激活数的 $m\times n$ 矩阵。
$B$ 是量化权重的 $n\times p$ 矩阵。
假设将 $A$ 的第 $j$ 行 $a_j$ 乘以 $B$、$b_k$，长度均为 $n$。量化的整数值和零分值为 $q_a$、$z_a$ 和 $q_b$、$z_b$。

\[a_j \cdot b_k = \sum_{i=0}^{n} a_{j}^{(i)} b_{k}^{(i)} = \sum_{i=0}^{n} (q_{a}^{(i)} - z_a) (q_{b}^{(i)} - z_b) = \sum_{i=0}^{n} q_{a}^{(i)} q_{b}^{(i)} - \sum_{i=0}^{n} q_{a}^{(i)} z_b - \sum_{i=0}^{n} q_{b}^{(i)} z_a + \sum_{i=0}^{n} z_a z_b\]

将

该 $\sum_{i=0}^{n} q_{a}^{(i)} q_{b}^{(i)}$ 字词是不可避免的，因为它计算输入值与权重值的点积。

$\sum_{i=0}^{n} q_{b}^{(i)} z_a$ 和 $\sum_{i=0}^{n} z_a z_b$ 条款由每次推理调用保持不变的常量组成，因此可以进行预先计算

$\sum_{i=0}^{n} q_{a}^{(i)} z_b$ 项在每次推断时都需要计算因为激活会改变每次推理。通过将权重设为对称，我们可以移除该字词的费用。

int8 量化运算符规范

下面，我们介绍了 int8 tflite 内核的量化要求：

ADD
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Input 1:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Output 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor

AVERAGE_POOL_2D
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Output 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  restriction: Input and outputs must all have same scale/zero_point

CONCATENATION
  Input ...:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Output 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  restriction: Input and outputs must all have same scale/zero_point

CONV_2D
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Input 1 (Weight):
    data_type  : int8
    range      : [-127, 127]
    granularity: per-axis (dim = 0)
    restriction: zero_point = 0
  Input 2 (Bias):
    data_type  : int32
    range      : [int32_min, int32_max]
    granularity: per-axis
    restriction: (scale, zero_point) = (input0_scale * input1_scale[...], 0)
  Output 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor

DEPTHWISE_CONV_2D
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Input 1 (Weight):
    data_type  : int8
    range      : [-127, 127]
    granularity: per-axis (dim = 3)
    restriction: zero_point = 0
  Input 2 (Bias):
    data_type  : int32
    range      : [int32_min, int32_max]
    granularity: per-axis
    restriction: (scale, zero_point) = (input0_scale * input1_scale[...], 0)
  Output 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor

FULLY_CONNECTED
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Input 1 (Weight):
    data_type  : int8
    range      : [-127, 127]
    granularity: per-axis (dim = 0)
    restriction: zero_point = 0
  Input 2 (Bias):
    data_type  : int32
    range      : [int32_min, int32_max]
    granularity: per-tensor
    restriction: (scale, zero_point) = (input0_scale * input1_scale[...], 0)
  Output 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor

L2_NORMALIZATION
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Output 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
    restriction: (scale, zero_point) = (1.0 / 128.0, 0)

LOGISTIC
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Output 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
    restriction: (scale, zero_point) = (1.0 / 256.0, -128)

MAX_POOL_2D
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Output 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  restriction: Input and outputs must all have same scale/zero_point

MUL
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Input 1:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Output 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor

RESHAPE
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Output 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  restriction: Input and outputs must all have same scale/zero_point

RESIZE_BILINEAR
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Output 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  restriction: Input and outputs must all have same scale/zero_point

SOFTMAX
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Output 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
    restriction: (scale, zero_point) = (1.0 / 256.0, -128)

SPACE_TO_DEPTH
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Output 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  restriction: Input and outputs must all have same scale/zero_point

TANH
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Output 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
    restriction: (scale, zero_point) = (1.0 / 128.0, 0)

PAD
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Output 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  restriction: Input and outputs must all have same scale/zero_point

GATHER
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Output 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  restriction: Input and outputs must all have same scale/zero_point

BATCH_TO_SPACE_ND
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Output 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  restriction: Input and outputs must all have same scale/zero_point

SPACE_TO_BATCH_ND
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Output 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  restriction: Input and outputs must all have same scale/zero_point

TRANSPOSE
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Output 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  restriction: Input and outputs must all have same scale/zero_point

MEAN
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Output 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor

SUB
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Input 1:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Output 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor

SUM
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Output 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor

SQUEEZE
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Output 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  restriction: Input and outputs must all have same scale/zero_point

LOG_SOFTMAX
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Output 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
    restriction: (scale, zero_point) = (16.0 / 256.0, 127)

MAXIMUM
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Output 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  restriction: Input and outputs must all have same scale/zero_point

ARG_MAX
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor

MINIMUM
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Output 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  restriction: Input and outputs must all have same scale/zero_point

LESS
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Input 1:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor

PADV2
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Output 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  restriction: Input and outputs must all have same scale/zero_point

GREATER
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Input 1:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor

GREATER_EQUAL
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Input 1:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor

LESS_EQUAL
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Input 1:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor

SLICE
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Output 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  restriction: Input and outputs must all have same scale/zero_point

EQUAL
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Input 1:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor

NOT_EQUAL
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Input 1:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor

SHAPE
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor

QUANTIZE (Requantization)
  Input 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor
  Output 0:
    data_type  : int8
    range      : [-128, 127]
    granularity: per-tensor

参考

arXiv:1712.05877