Quantizers

Top-level API

class aimet_torch.v2.quantization.affine.quantizer.QuantizerBase[source]

Quantizer base class

allow_overwrite(mode)[source]

Set allow_overwite flag

abstract compute_encodings()[source]

Observe inputs and update quantization parameters based on the input statistics.

abstract get_encoding()[source]

Return the quantizer’s encodings as an EncodingBase object

Return type:

Optional[EncodingBase]

abstract get_legacy_encodings()[source]

Returns a list of encodings, each represented as a List of Dicts

Return type:

Optional[List[Dict]]

is_initialized()[source]

Returns true if the quantization parameters are initialized.

Return type:

bool

register_quantization_parameter(name, param)[source]

Register quantization parameter.

abstract set_legacy_encodings(encodings)[source]

Set encodings represented in the same format as the output of get_legacy_encodings.

class aimet_torch.v2.quantization.affine.quantizer.QuantizeDequantize(shape, bitwidth, symmetric, encoding_analyzer=None, block_size=None)[source]

Applies fake-quantization by quantizing and dequantizing the input.

Precisely,

\[out = (\overline{input} + offset) * scale\]

where

\[\overline{input} = clamp\left(\left\lceil\frac{input}{scale}\right\rfloor - offset, qmin, qmax\right)\]

and \(scale\) and \(offset\) are derived from learnable parameters \(\theta_{min}\) and \(\theta_{max}\).

If block size \(B = \begin{pmatrix} B_0 & B_1 & \cdots & B_{D-1} \end{pmatrix}\) is specified, this equation will be further generalized as

\[ \begin{align}\begin{aligned}\begin{split}out_{j_0 \cdots j_{D-1}} &= (\overline{input}_{j_0 \cdots j_{D-1}} + offset_{i_0 \cdots i_{D-1}}) * scale_{i_0 \cdots i_{D-1}}\\ \overline{input}_{j_0 \cdots j_{D-1}} &= clamp\left( \left\lceil\frac{input_{j_0 \cdots j_{D-1}}}{scale_{i_0 \cdots i_{D-1}}}\right\rfloor - offset_{i_0 \cdots i_{D-1}}, qmin, qmax\right)\\\end{split}\\\text{where} \quad \forall_{0 \leq d < D} \quad i_d = \left\lfloor \frac{j_d}{B_d} \right\rfloor\end{aligned}\end{align} \]
Parameters:
  • shape (tuple) – Shape of the quantization parameters

  • bitwidth (int) – Quantization bitwidth

  • symmetric (bool) – If True, performs symmetric quantization; otherwise, performs asymmetric quantization

  • encoding_analyzer (EncodingAnalyzer, optional) – Encoding analyzer for calibrating quantization encodings (default: absolute min-max encoding analyzer)

  • block_size (Tuple[int, ...], optional) – Block size

Variables:
  • min (Tensor) – \(\theta_{min}\) from which scale and offset will be derived.

  • max (Tensor) – \(\theta_{max}\) from which scale and offset will be derived.

Note

QuantizeDequantize cannot run forward() until min and max are properly initialized, which can be done based on input statistics using compute_encodings() or by manually assigning a new value to min and max. See the examples below.

Examples

>>> import aimet_torch.v2.quantization as Q
>>> input = torch.randn(5, 10)
>>> qdq = Q.affine.QuantizeDequantize(shape=(5, 2), bitwidth=8, symmetric=False, block_size=(1, 5))
>>> qdq.is_initialized()
False
>>> with qdq.compute_encodings():
...     _ = qdq(input)
...
>>> qdq.is_initialized()
True
>>> qdq(input)
DequantizedTensor([[-0.2771,  0.3038,  1.0819,  0.9700,  0.9487, -0.1307,
                    -1.7894, -0.1709, -0.2212,  0.7741],
                   [-1.0295, -1.2265, -1.0295,  1.0564,  0.6177, -1.0386,
                    -0.0176, -2.6054,  1.8836, -0.1232],
                   [-0.8229,  0.5540,  0.3992, -0.2363,  1.2546, -1.0036,
                     0.2355,  0.1741,  1.6079,  0.6247],
                   [-1.0115,  1.2458,  0.9157, -1.4694, -0.0639, -0.2568,
                     0.0680,  1.6695,  0.7932, -0.1889],
                   [ 0.0158,  0.5695,  0.5220,  0.1977, -1.4475, -0.0424,
                    -1.1128, -0.8796, -0.1060,  1.5897]],
                  grad_fn=<AliasBackward0>)
>>> import aimet_torch.v2.quantization as Q
>>> input = torch.randn(5, 10)
>>> qdq = Q.affine.QuantizeDequantize(shape=(5, 2), bitwidth=8, symmetric=False, block_size=(1, 5))
>>> qdq.is_initialized()
False
>>> qdq.min = torch.nn.Parameter(-torch.ones_like(qdq.min))
>>> qdq.max = torch.nn.Parameter(torch.ones_like(qdq.max))
>>> qdq.is_initialized()
True
>>> qdq(input)
DequantizedTensor([[-0.6196, -0.9961,  0.0549, -0.6431,  1.0039, -0.8706,
                     1.0039,  0.4706, -0.2353,  0.8078],
                   [ 0.3451, -0.1176, -0.9961, -0.4549, -0.0549, -0.0471,
                    -0.5255, -0.2353,  1.0039, -0.9961],
                   [-0.4157,  0.0784,  0.5333,  0.1647, -0.9961, -0.9961,
                    -0.2118, -0.2196,  0.9176,  0.9490],
                   [ 1.0039, -0.7765,  0.4784, -0.8706,  1.0039,  0.6039,
                    -0.4157, -0.2118, -0.9961,  0.3137],
                   [ 1.0039,  0.3216, -0.2353, -0.7765, -0.9961,  0.8000,
                     1.0039,  0.4157,  0.4392,  0.4863]],
                  grad_fn=<AliasBackward0>)
forward(input)[source]

Quantizes and dequantizes the input tensor

Return type:

DequantizedTensor

Parameters:

input (torch.Tensor) – Input to quantize and dequantize

Returns:

Quantize-dequantized output

class aimet_torch.v2.quantization.affine.quantizer.Quantize(shape, bitwidth, symmetric, encoding_analyzer=None, block_size=None)[source]

Applies quantization to the input.

Precisely,

\[out = clamp\left(\left\lceil\frac{input}{scale}\right\rfloor - offset, qmin, qmax\right)\]

where \(scale\) and \(offset\) are derived from learnable parameters \(\theta_{min}\) and \(\theta_{max}\).

If block size \(B = \begin{pmatrix} B_0 & B_1 & \cdots & B_{D-1} \end{pmatrix}\) is specified, this equation will be further generalized as

\[ \begin{align}\begin{aligned}\begin{split}out_{j_0 \cdots j_{D-1}} & = clamp\left( \left\lceil\frac{input_{j_0 \cdots j_{D-1}}}{scale_{i_0 \cdots i_{D-1}}}\right\rfloor - offset_{i_0 \cdots i_{D-1}}, qmin, qmax\right)\\\end{split}\\\text{where} \quad \forall_{0 \leq d < D} \quad i_d = \left\lfloor \frac{j_d}{B_d} \right\rfloor\end{aligned}\end{align} \]
Parameters:
  • shape (tuple) – Shape of the quantization parameters

  • bitwidth (int) – Quantization bitwidth

  • symmetric (bool) – If True, performs symmetric quantization; otherwise, performs asymmetric quantization

  • encoding_analyzer (EncodingAnalyzer, optional) – Encoding analyzer for calibrating quantization encodings (default: absolute min-max encoding analyzer)

  • block_size (Tuple[int, ...], optional) – Block size

Variables:
  • min (Tensor) – \(\theta_{min}\) from which scale and offset will be derived.

  • max (Tensor) – \(\theta_{max}\) from which scale and offset will be derived.

Note

Quantize cannot run forward() until min and max are properly initialized, which can be done based on input statistics using compute_encodings() or by manually assigning a new value to min and max. See the examples below.

Examples

>>> import aimet_torch.v2.quantization as Q
>>> input = torch.randn(5, 10)
>>> q = Q.affine.Quantize(shape=(5, 1), bitwidth=8, symmetric=False, block_size=(1, 5))
>>> q.is_initialized()
False
>>> with q.compute_encodings():
...     _ = q(input)
...
>>> q.is_initialized()
True
>>> q(input)
QuantizedTensor([[129.,  64., 255., 122.,   0., 192., 106.,  94., 255.,   0.],
                 [  0., 145., 181., 255., 144., 255., 194.,   0.,  74.,  86.],
                 [122.,   0., 255., 150.,  33., 103., 103.,   0.,  37., 255.],
                 [255., 111., 237., 218.,   0.,  49., 155., 255.,   0., 179.],
                 [  0.,  66., 255.,  89., 110.,  17.,  36.,  83., 255.,   0.]],
                grad_fn=<AliasBackward0>)
>>> import aimet_torch.v2.quantization as Q
>>> input = torch.randn(5, 10)
>>> q = Q.affine.Quantize(shape=(5, 1), bitwidth=8, symmetric=False, block_size=(1, 5))
>>> q.is_initialized()
False
>>> q.min = torch.nn.Parameter(-torch.ones_like(q.min))
>>> q.max = torch.nn.Parameter(torch.ones_like(q.max))
>>> q.is_initialized()
True
>>> q(input)
QuantizedTensor([[187., 186., 131.,   0., 203.,  64.,  80.,   0., 143., 152.],
                 [ 16.,   0., 255.,   0.,   0., 150.,   0., 255.,  32., 255.],
                 [255., 226.,   0., 255.,  55., 172.,   0., 255., 145., 255.],
                 [207., 146., 216., 238.,   0.,   0., 141., 178., 255., 188.],
                 [ 63.,  59.,  19., 162.,  30., 255., 109., 255.,   0., 255.]],
                grad_fn=<AliasBackward0>)
forward(input)[source]

Quantizes the input tensor

Return type:

QuantizedTensor

Parameters:

input (torch.Tensor) – Input to quantize

Returns:

Quantized output