terminator.models.layers.gvp.GVP

class terminator.models.layers.gvp.GVP(in_dims, out_dims, h_dim=None, activations=(<function relu>, <built-in method sigmoid of type object>), vector_gate=False)

Bases: Module

Geometric Vector Perceptron. See manuscript and README.md for more details.

Parameters:

• in_dims – tuple (n_scalar, n_vector)

• out_dims – tuple (n_scalar, n_vector)

• h_dim – intermediate number of vector channels, optional

• activations – tuple of functions (scalar_act, vector_act)

• vector_gate – whether to use vector gating. (vector_act will be used as sigma^+ in vector gating if True)

__init__(in_dims, out_dims, h_dim=None, activations=(<function relu>, <built-in method sigmoid of type object>), vector_gate=False)

Initializes internal Module state, shared by both nn.Module and ScriptModule.

Methods

__init__(in_dims, out_dims[, h_dim, ...])	Initializes internal Module state, shared by both nn.Module and ScriptModule.
forward(x)	param x: tuple (s, V) of torch.Tensor,

Attributes

T_destination	alias of TypeVar('T_destination', bound=Mapping[str, Tensor])
dump_patches	This allows better BC support for load_state_dict().

forward(x)

Parameters:

x – tuple (s, V) of torch.Tensor, or (if vectors_in is 0), a single torch.Tensor

Returns:

tuple (s, V) of torch.Tensor, or (if vectors_out is 0), a single torch.Tensor