tltorch.factorized_layers.TCL

class tltorch.factorized_layers.TCL(input_shape, rank, verbose=0, bias=False, device=None, dtype=None, **kwargs)[source]

Tensor Contraction Layer [1]

Parameters:

input_sizeint iterable

shape of the input, excluding batch size

rankint list or int

rank of the TCL, will also be the output-shape (excluding batch-size) if int, the same rank will be used for all dimensions

verboseint, default is 1

level of verbosity

References

[1]

J. Kossaifi, A. Khanna, Z. Lipton, T. Furlanello and A. Anandkumar, “Tensor Contraction Layers for Parsimonious Deep Nets,” 2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), Honolulu, HI, 2017, pp. 1940-1946, doi: 10.1109/CVPRW.2017.243.

Attributes:

factors

Methods

forward(x)	Performs a forward pass
reset_parameters()	Sets the parameters' values randomly

forward(x)[source]

Performs a forward pass

reset_parameters()[source]

Sets the parameters’ values randomly