assembly.Simulator

class assembly.Simulator(model, epoch_size=10, env_suffix='')

Bases: object

Train and simulate Computation with Assemblies.

Parameters

modelAreaInterface

A NN model, consisting of one or more areas.

epoch_sizeint, optional

Defines the number of simulations to run for each trial sample. Each trial sample represent a complete epoch. Default: 10

Methods

associate_benchmark(x_samples[, learned])	Measure associate operation overlap between projected assemblies from two (or more) parent areas.
simulate(x_samples)	Train and simulate the model with the input x_samples data.

associate_benchmark(x_samples, learned=False)

Measure associate operation overlap between projected assemblies from two (or more) parent areas. Each assembly is projected individually.

Parameters

x_sampleslist of tuple of torch.Tensor

The input stimuli samples list. In this case, each entry must be a pair of vectors.

learnedbool, optional

Have the model areas been associated (True) or not? This flag is used in the plots title only. Default: False

simulate(x_samples)

Train and simulate the model with the input x_samples data.

Parameters

x_sampleslist

The input stimuli samples list. Each entry can be either a vector tensor or, in case of AreaStack, a pair of tensors.