composer.algorithms.factorize.factorize_modules#

composer.algorithms.factorize.factorize_modules

Functions

`cast`	Cast a value to a type.
`factorize_conv2d`	Approximates a KxK convolution by factorizing it into a KxK convolution with fewer channels followed by a 1x1 convolution.
`factorize_matrix`	Approximates a matrix by factorizing it into a product of two smaller matrices.
`factorizing_could_speedup`	Whether factorizing a module a given amount could possibly yield a benefit.

Classes

`FactorizedConv2d`	Factorized replacement for `torch.nn.Conv2d`.
`FactorizedLinear`	Factorized replacement for `torch.nn.Linear`.
`LowRankSolution`	Bundles tensors used by a factorized linear operator.

Attributes

Optional
Tuple
Union
annotations

class composer.algorithms.factorize.factorize_modules.FactorizedConv2d(in_channels, out_channels, kernel_size, latent_channels=0.25, **kwargs)[source]#

Bases: composer.algorithms.factorize.factorize_modules._FactorizedModule

Factorized replacement for torch.nn.Conv2d.

Splits the conv2d operation into two smaller conv2d operations, which are executed sequentially with no nonlinearity in between. This first conv2d can be thought of as projecting the feature maps into a lower-dimensional space, similar to PCA. The second produces outputs of the same shape as the un-factorized version based on the embeddings within this lower-dimensional space. Note that “dimensionality” here refers to the number of channels, not the spatial extent or tensor rank.

The first conv2d has a kernel size of kernel_size, while the second one always has a kernel size of 1x1. For large kernel sizes, the lower-dimensional space can be nearly as large as min(in_channels, out_channels) and still yield a reduction in multiply-add operations. For kernels sizes of 1x1, the breakeven point is a 2x reduction in channel count, similar to FactorizedLinear.

See factorize_conv2d() for more details.

Parameters

in_channels (int) – number of channels in the input image
out_channels (int) – number of channels produced by the convolution
kernel_size (int or tuple, optional) – size of the convolving kernel
latent_channels (int or float, optional) – number of channels in the latent representation produced by the first small convolution. Can be specified as either an integer > 1 or as float within [0, 1). In the latter case, the value is interpreted as a fraction of min(in_features, out_features) for each linear module, and is converted to the equivalent integer value, with a minimum of 1.
**kwargs – other arguments to torch.nn.Conv2d are supported and will be used with the first of the two smaller Conv2d operations. However, groups > 1 and dilation > 1 are not currently supported.

Raises

ValueError – If latent_channels is not small enough for factorization to reduce the number of multiply-add operations. In this regime, factorization is both slower and less expressive than a non-factorized operation. Setting latent_features to max_allowed_latent_channels() or a smaller value is sufficient to avoid this.

property in_channels#: See torch.nn.Conv2d.

property latent_channels#: The number of of output channels for the first convolution, which is also the number of input channels for the second convolution.

static max_allowed_latent_features(in_features, out_features, kernel_size)[source]#

Returns the largest latent channel count that reduces the number of multiply-adds.

Parameters

in_channels – number of channels in the input image
out_channels – number of channels produced by the convolution
kernel_size – size of the convolving kernel

Returns

latent_channels – the largest allowable number of latent channels

property out_channels#: See torch.nn.Conv2d.

class composer.algorithms.factorize.factorize_modules.FactorizedLinear(in_features, out_features, bias=True, latent_features=0.25)[source]#

Bases: composer.algorithms.factorize.factorize_modules._FactorizedModule

Factorized replacement for torch.nn.Linear.

Splits the linear operation into two smaller linear operations, which are executed sequentially with no nonlinearity in between. This first linear operation can be thought of as projecting the inputs into a lower-dimensional space, similar to PCA. The second produces outputs of the same shape as the un-factorized version based on the embeddings within this lower-dimensional space.

If the lower-dimensional space is less than half the size of the smaller of the input and output dimensionality, this factorization can reduce the number of multiply-adds necessary to compute the output. However, because larger matrix products tend to utilize the hardware better, it may take a reduction of more than 2x to get a speedup in practice.

See factorize_matrix() for more details.

Parameters

in_features (int) – size of each input sample
out_features (int) – size of each output sample
bias (bool) – If set to False, the layer will not learn an additive bias.
latent_features (int or float, optional) – size of the latent space. Can be specified as either an integer > 1 or as a float within [0, 0.5). In the latter case, the value is interpreted as a fraction of min(in_features, out_features), and is converted to the equivalent integer value, with a minimum of 1.

Raises

ValueError – If latent_features is not small enough for factorization to reduce the number of multiply-add operations. In this regime, factorization is both slower and less expressive than a non-factorized operation. Setting latent_features < min(in_features, out_features) / 2 or using max_allowed_latent_features() is sufficient to avoid this.

property in_features#: See torch.nn.Linear.

property latent_features#: The dimensionality of the space into which the input is projected by the first matrix in the factorization.

static max_allowed_latent_channels(in_features, out_features)[source]#

Returns the largest latent feature count that reduces the number of multiply-adds.

Parameters

in_features – size of each input sample
out_features – size of each output sample

Returns

latent_features – the largest allowable number of latent features

property out_features#: See torch.nn.Linear.

composer.algorithms.factorize.factorize_modules.factorizing_could_speedup(module, latent_size)[source]#

Whether factorizing a module a given amount could possibly yield a benefit.

This computation is based on the number of multiply-add operations involved in the module’s current forward pass versus the number that would be involved if it were factorized into two modules using the specified latent size. The operations are assumed to be dense and of the same data type in all cases.

Note that this function returning true does not guarantee a wall-clock speedup, since splitting one operation into two involves more data movement and more per-op overhead.

Parameters

module (Module) – a Conv2d, Linear, FactorizedConv2d, or FactorizedLinear.
latent_size (int or float) – number of channels (for convolution) or features (for linear) in the latent representation. Can be specified as either an integer > 1 or as float within [0, 1). In the latter case, the value is interpreted as a fraction of min(in_features, out_features) for a linear module or min(in_channels, out_channels) for a convolution.

Returns

could_speedup – A bool indicating whether the provided amount of factorization could accelerate the provided module. If module is not one of the allowed types, always returns False, since there is no supported way to factorize that module.