jittor_geometric.nn.dense

Dense layers and utilities.

class jittor_geometric.nn.dense.MergeLayer(input_dim1, input_dim2, hidden_dim, output_dim)

Bases: Module

Parameters:

• input_dim1 (int)

• input_dim2 (int)

• hidden_dim (int)

• output_dim (int)

__init__(input_dim1, input_dim2, hidden_dim, output_dim)

Merge Layer to merge two inputs via: input_dim1 + input_dim2 -> hidden_dim -> output_dim. :type input_dim1: int :param input_dim1: int, dimension of first input :type input_dim2: int :param input_dim2: int, dimension of the second input :type hidden_dim: int :param hidden_dim: int, hidden dimension :type output_dim: int :param output_dim: int, dimension of the output

execute(input_1, input_2)

merge and project the inputs :type input_1: Var :param input_1: Var, shape (, input_dim1) :type input_2: :sphinx_autodoc_typehints_type:`:py:class:`~jittor.jittor_core.Var`` :param input_2: Var, shape (, input_dim2) :return:

class jittor_geometric.nn.dense.TimeEncoder(time_dim, parameter_requires_grad=True)

Bases: Module

Parameters:

• time_dim (int)

• parameter_requires_grad (bool)

__init__(time_dim, parameter_requires_grad=True)

Time encoder. :type time_dim: int :param time_dim: int, dimension of time encodings :type parameter_requires_grad: bool :param parameter_requires_grad: boolean, whether the parameter in TimeEncoder needs gradient

execute(timestamps)

compute time encodings of time in timestamps :type timestamps: Var :param timestamps: Var, shape (batch_size, seq_len) :return:

class jittor_geometric.nn.dense.MLP(num_layers, input_dim, hidden_dim, output_dim, act_type='PReLU', dropout=0.1, use_act=True, skip_connection=True, pinit=0.15)

Bases: Module

Parameters:

• num_layers (int)

• input_dim (int)

• hidden_dim (int)

• output_dim (int)

• dropout (float)

• use_act (bool)

• skip_connection (bool)

• pinit (float)

__init__(num_layers, input_dim, hidden_dim, output_dim, act_type='PReLU', dropout=0.1, use_act=True, skip_connection=True, pinit=0.15)

Parameters:

• num_layers (int)

• input_dim (int)

• hidden_dim (int)

• output_dim (int)

• dropout (float)

• use_act (bool)

• skip_connection (bool)

• pinit (float)

execute(input)

Executes the module computation.

Raises NotImplementedError if the subclass does not override the method.

Parameters:

input (Var)