TensorFlow Lite 8 位元量化規格

以下文件概述 TensorFlow Lite 8 位元量化配置的規格。旨在協助硬體開發人員為使用量化 TensorFlow Lite 模型進行推論提供硬體支援。

規格摘要

我們提供規格，且唯有遵循規格，我們才能保證行為。我們也瞭解，不同的硬體可能會有偏好和限制，進而導致實作規格時產生些微偏差，導致實作結果並非位元精確。雖然這在大多數情況下可能可以接受 (我們會提供一套測試，據我們所知，這套測試包含我們從多個模型收集而來的每個運算容許值)，但機器學習 (以及最常見的深度學習) 的本質使我們無法提供任何明確的保證。

8 位元量化使用下列公式來近似浮點值。

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

每個軸 (又稱為 Conv 運算中的每個通道) 或每個張量權重由 int8 二補數值表示，範圍為 [-127, 127]，零點等於 0。每個張量啟動/輸入由 int8 二補數值表示，範圍為 [-128, 127]，零點範圍為 [-128, 127]。

以下文件記錄了特定運算的其他例外情況。

帶正負號整數與不帶正負號整數

TensorFlow Lite 量化將優先考量 int8 量化的工具和核心，以進行 8 位元量化。這是為了方便將零點等於 0 的對稱量化表示出來。此外，許多後端都針對 int8xint8 累加進行了額外的最佳化。

每個軸與每個張量

每個張量量化表示每個完整張量都會有一個縮放比例和/或零點。每個軸量化表示每個切片在 quantized_dimension 中都會有一個縮放比例和/或 zero_point。quantized_dimension 指定張量形狀的維度，縮放比例和零點會對應至該維度。例如，張量 t，具有 dims=[4, 3, 2, 1] 和量化參數：scale=[1.0, 2.0, 3.0]、zero_point=[1, 2, 3]、quantization_dimension=1，將會在 t 的第二個維度上進行量化

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 是迴旋積權重的 output_channel，但理論上它可以是與核心實作中的每個點積對應的維度，進而在不影響效能的情況下，實現更高的量化精細度。這對準確性有很大的改進。

TFLite 對越來越多的運算提供每個軸的支援。在本文件撰寫時，支援 Conv2d 和 DepthwiseConv2d。

對稱與非對稱

啟動是非對稱的：它們的零點可以位於帶正負號 int8 範圍 [-128, 127] 內的任何位置。許多啟動本質上是非對稱的，而零點是一種相對便宜的方式，可以有效地獲得額外的二進位位元精確度。由於啟動只會乘以常數權重，因此常數零點值可以大幅最佳化。

權重是對稱的：強制零點等於 0。權重值會乘以動態輸入和啟動值。這表示權重的零點與啟動值的乘法運算會產生無法避免的執行階段成本。透過強制零點為 0，我們可以避免此成本。

數學說明：這與 arXiv:1712.05877 中第 2.3 節類似，但不同之處在於我們允許縮放比例值為每個軸。這可以很容易地推廣，如下所示

\(A\) 是 \(m \times n\) 量化啟動矩陣。
\(B\) 是 \(n \times p\) 量化權重矩陣。
考慮將 \(A\) 的第 \(j\) 列 \(a_j\) 乘以 \(B\) 的第 \(k\) 行 \(b_k\) (長度皆為 \(n\))。量化整數值和零點值分別為 \(q_a\)、\(z_a\) 和 \(q_b\)、\(z_b\)。

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

\(\sum_{i=0}^{n} q_{a}^{(i)} q_{b}^{(i)}\) 項是不可避免的，因為它正在執行輸入值和權重值的點積。

\(\sum_{i=0}^{n} q_{b}^{(i)} z_a\) 和 \(\sum_{i=0}^{n} z_a z_b\) 項由常數組成，這些常數在每次推論調用中都保持不變，因此可以預先計算。

\(\sum_{i=0}^{n} q_{a}^{(i)} z_b\) 項需要在每次推論時計算，因為啟動在每次推論時都會變更。透過強制權重為對稱，我們可以消除此項的成本。

int8 量化運算子規格

以下說明我們的 int8 tflite 核心的量化需求

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

參考資料

arXiv:1712.05877