deepmd.pt.model.network.mlp

deepmd.pt.model.network.mlp#

Attributes#

device

MLP_

EmbeddingNet

FittingNet

Classes#

Identity

MLPLayer

MLP

Native representation of a neural network.

NetworkCollection

PyTorch implementation of NetworkCollection.

Functions#

empty_t(shape, precision)

Module Contents#

deepmd.pt.model.network.mlp.device[source]#
deepmd.pt.model.network.mlp.empty_t(shape, precision)[source]#
class deepmd.pt.model.network.mlp.Identity[source]#

Bases: torch.nn.Module

forward(xx: torch.Tensor) → torch.Tensor[source]#

The Identity operation layer.

serialize() → dict[source]#
classmethod deserialize(data: dict) → Identity[source]#
class deepmd.pt.model.network.mlp.MLPLayer(num_in, num_out, bias: bool = True, use_timestep: bool = False, activation_function: str | None = None, resnet: bool = False, bavg: float = 0.0, stddev: float = 1.0, precision: str = DEFAULT_PRECISION, init: str = 'default', seed: int | list[int] | None = None)[source]#

Bases: torch.nn.Module

use_timestep[source]#
num_in[source]#
num_out[source]#
activate_name = None[source]#
activate[source]#
precision = 'float64'[source]#
prec[source]#
matrix[source]#
resnet = False[source]#
check_type_consistency() → None[source]#
dim_in() → int[source]#
dim_out() → int[source]#
_default_normal_init(bavg: float = 0.0, stddev: float = 1.0, generator: torch.Generator | None = None) → None[source]#
_trunc_normal_init(scale=1.0, generator: torch.Generator | None = None) → None[source]#
_glorot_uniform_init(generator: torch.Generator | None = None) → None[source]#
_zero_init(use_bias=True) → None[source]#
_normal_init(generator: torch.Generator | None = None) → None[source]#
forward(xx: torch.Tensor) → torch.Tensor[source]#

One MLP layer used by DP model.

Parameters:
xxtorch.Tensor

The input.

Returns:
yy: torch.Tensor

The output.

serialize() → dict[source]#

Serialize the layer to a dict.

Returns:
dict

The serialized layer.

classmethod deserialize(data: dict) → MLPLayer[source]#

Deserialize the layer from a dict.

Parameters:
datadict

The dict to deserialize from.

deepmd.pt.model.network.mlp.MLP_[source]#
class deepmd.pt.model.network.mlp.MLP(*args, **kwargs)[source]#

Bases: MLP_

Native representation of a neural network.

Parameters:
layerslist[NativeLayer], optional

The layers of the network.

layers[source]#
forward[source]#
deepmd.pt.model.network.mlp.EmbeddingNet[source]#
deepmd.pt.model.network.mlp.FittingNet[source]#
class deepmd.pt.model.network.mlp.NetworkCollection(*args, **kwargs)[source]#

Bases: deepmd.dpmodel.utils.NetworkCollection, torch.nn.Module

PyTorch implementation of NetworkCollection.

NETWORK_TYPE_MAP: ClassVar[dict[str, type]][source]#
Contents