nlp_architect.models.temporal_convolutional_network.TCN

class nlp_architect.models.temporal_convolutional_network.TCN(max_len, n_features_in, hidden_sizes, kernel_size=7, dropout=0.2)[source]

This class defines core TCN architecture. This is only the base class, training strategy is not implemented.

__init__(max_len, n_features_in, hidden_sizes, kernel_size=7, dropout=0.2)[source]
To use this class,
  1. Inherit this class

  2. Define the training losses in build_train_graph()

  3. Define the training strategy in run()

  4. After the inherited class object is initialized, call build_train_graph followed by run

Parameters
  • max_len – Maximum length of sequence

  • n_features_in – Number of input features (dimensions)

  • hidden_sizes – Number of hidden sizes in each layer of TCN (same for all layers)

  • kernel_size – Kernel size of convolution filter (same for all layers)

  • dropout – Dropout, fraction of activations to drop

Methods

__init__(max_len, n_features_in, hidden_sizes)

To use this class,

build_network_graph(x[, last_timepoint])

Given the input placeholder x, build the entire TCN graph :param x: Input placeholder :param last_timepoint: Whether or not to select only the last timepoint to output

build_train_graph(*args, **kwargs)

Placeholder for defining training losses and metrics

calculate_receptive_field()

Returns:

run(*args, **kwargs)

Placeholder for defining training strategy

build_network_graph(x, last_timepoint=False)[source]

Given the input placeholder x, build the entire TCN graph :param x: Input placeholder :param last_timepoint: Whether or not to select only the last timepoint to output

Returns

output of the TCN

build_train_graph(*args, **kwargs)[source]

Placeholder for defining training losses and metrics

calculate_receptive_field()[source]

Returns:

run(*args, **kwargs)[source]

Placeholder for defining training strategy