deepmd.pt.loss

Submodules

• deepmd.pt.loss.denoise
• deepmd.pt.loss.dos
• deepmd.pt.loss.ener
• deepmd.pt.loss.ener_spin
• deepmd.pt.loss.loss
• deepmd.pt.loss.tensor

Package Contents

Classes

DenoiseLoss	Base class for all neural network modules.
DOSLoss	Base class for all neural network modules.
EnergyStdLoss	Base class for all neural network modules.
EnergySpinLoss	Base class for all neural network modules.
TaskLoss	Base class for all neural network modules.
TensorLoss	Base class for all neural network modules.

class deepmd.pt.loss.DenoiseLoss(ntypes, masked_token_loss=1.0, masked_coord_loss=1.0, norm_loss=0.01, use_l1=True, beta=1.0, mask_loss_coord=True, mask_loss_token=True, **kwargs)

Bases: deepmd.pt.loss.loss.TaskLoss

Base class for all neural network modules.

Your models should also subclass this class.

Modules can also contain other Modules, allowing to nest them in a tree structure. You can assign the submodules as regular attributes:

import torch.nn as nn
import torch.nn.functional as F

class Model(nn.Module):
    def __init__(self):
        super().__init__()
        self.conv1 = nn.Conv2d(1, 20, 5)
        self.conv2 = nn.Conv2d(20, 20, 5)

    def forward(self, x):
        x = F.relu(self.conv1(x))
        return F.relu(self.conv2(x))

Submodules assigned in this way will be registered, and will have their parameters converted too when you call to(), etc.

Note

As per the example above, an __init__() call to the parent class must be made before assignment on the child.

Variables:

training (bool) – Boolean represents whether this module is in training or evaluation mode.

forward(model_pred, label, natoms, learning_rate, mae=False)

Return loss on coord and type denoise.

Returns:

• loss: Loss to minimize.

class deepmd.pt.loss.DOSLoss(starter_learning_rate: float, numb_dos: int, start_pref_dos: float = 1.0, limit_pref_dos: float = 1.0, start_pref_cdf: float = 1000, limit_pref_cdf: float = 1.0, start_pref_ados: float = 0.0, limit_pref_ados: float = 0.0, start_pref_acdf: float = 0.0, limit_pref_acdf: float = 0.0, inference=False, **kwargs)

Bases: deepmd.pt.loss.loss.TaskLoss

Base class for all neural network modules.

Your models should also subclass this class.

Modules can also contain other Modules, allowing to nest them in a tree structure. You can assign the submodules as regular attributes:

import torch.nn as nn
import torch.nn.functional as F

class Model(nn.Module):
    def __init__(self):
        super().__init__()
        self.conv1 = nn.Conv2d(1, 20, 5)
        self.conv2 = nn.Conv2d(20, 20, 5)

    def forward(self, x):
        x = F.relu(self.conv1(x))
        return F.relu(self.conv2(x))

Submodules assigned in this way will be registered, and will have their parameters converted too when you call to(), etc.

Note

As per the example above, an __init__() call to the parent class must be made before assignment on the child.

Variables:

training (bool) – Boolean represents whether this module is in training or evaluation mode.

property label_requirement: List[deepmd.utils.data.DataRequirementItem]

Return data label requirements needed for this loss calculation.

forward(input_dict, model, label, natoms, learning_rate=0.0, mae=False)

Return loss on local and global tensors.

Parameters:

input_dictdict[str, torch.Tensor]

Model inputs.

modeltorch.nn.Module

Model to be used to output the predictions.

labeldict[str, torch.Tensor]

Labels.

natomsint

The local atom number.

Returns:

model_pred: dict[str, torch.Tensor]

Model predictions.

loss: torch.Tensor

Loss for model to minimize.

more_loss: dict[str, torch.Tensor]

Other losses for display.

class deepmd.pt.loss.EnergyStdLoss(starter_learning_rate=1.0, start_pref_e=0.0, limit_pref_e=0.0, start_pref_f=0.0, limit_pref_f=0.0, start_pref_v=0.0, limit_pref_v=0.0, start_pref_ae: float = 0.0, limit_pref_ae: float = 0.0, start_pref_pf: float = 0.0, limit_pref_pf: float = 0.0, use_l1_all: bool = False, inference=False, **kwargs)

Bases: deepmd.pt.loss.loss.TaskLoss

Base class for all neural network modules.

Your models should also subclass this class.

Modules can also contain other Modules, allowing to nest them in a tree structure. You can assign the submodules as regular attributes:

import torch.nn as nn
import torch.nn.functional as F

class Model(nn.Module):
    def __init__(self):
        super().__init__()
        self.conv1 = nn.Conv2d(1, 20, 5)
        self.conv2 = nn.Conv2d(20, 20, 5)

    def forward(self, x):
        x = F.relu(self.conv1(x))
        return F.relu(self.conv2(x))

Submodules assigned in this way will be registered, and will have their parameters converted too when you call to(), etc.

Note

As per the example above, an __init__() call to the parent class must be made before assignment on the child.

Variables:

training (bool) – Boolean represents whether this module is in training or evaluation mode.

property label_requirement: List[deepmd.utils.data.DataRequirementItem]

Return data label requirements needed for this loss calculation.

forward(input_dict, model, label, natoms, learning_rate, mae=False)

Return loss on energy and force.

Parameters:

input_dictdict[str, torch.Tensor]

Model inputs.

modeltorch.nn.Module

Model to be used to output the predictions.

labeldict[str, torch.Tensor]

Labels.

natomsint

The local atom number.

Returns:

model_pred: dict[str, torch.Tensor]

Model predictions.

loss: torch.Tensor

Loss for model to minimize.

more_loss: dict[str, torch.Tensor]

Other losses for display.

class deepmd.pt.loss.EnergySpinLoss(starter_learning_rate=1.0, start_pref_e=0.0, limit_pref_e=0.0, start_pref_fr=0.0, limit_pref_fr=0.0, start_pref_fm=0.0, limit_pref_fm=0.0, start_pref_v=0.0, limit_pref_v=0.0, start_pref_ae: float = 0.0, limit_pref_ae: float = 0.0, start_pref_pf: float = 0.0, limit_pref_pf: float = 0.0, use_l1_all: bool = False, inference=False, **kwargs)

Bases: deepmd.pt.loss.loss.TaskLoss

Base class for all neural network modules.

Your models should also subclass this class.

Modules can also contain other Modules, allowing to nest them in a tree structure. You can assign the submodules as regular attributes:

import torch.nn as nn
import torch.nn.functional as F

class Model(nn.Module):
    def __init__(self):
        super().__init__()
        self.conv1 = nn.Conv2d(1, 20, 5)
        self.conv2 = nn.Conv2d(20, 20, 5)

    def forward(self, x):
        x = F.relu(self.conv1(x))
        return F.relu(self.conv2(x))

Submodules assigned in this way will be registered, and will have their parameters converted too when you call to(), etc.

Note

As per the example above, an __init__() call to the parent class must be made before assignment on the child.

Variables:

training (bool) – Boolean represents whether this module is in training or evaluation mode.

property label_requirement: List[deepmd.utils.data.DataRequirementItem]

Return data label requirements needed for this loss calculation.

forward(input_dict, model, label, natoms, learning_rate, mae=False)

Return energy loss with magnetic labels.

Parameters:

input_dictdict[str, torch.Tensor]

Model inputs.

modeltorch.nn.Module

Model to be used to output the predictions.

labeldict[str, torch.Tensor]

Labels.

natomsint

The local atom number.

Returns:

model_pred: dict[str, torch.Tensor]

Model predictions.

loss: torch.Tensor

Loss for model to minimize.

more_loss: dict[str, torch.Tensor]

Other losses for display.

class deepmd.pt.loss.TaskLoss(**kwargs)

Bases: torch.nn.Module, abc.ABC

Base class for all neural network modules.

Your models should also subclass this class.

Modules can also contain other Modules, allowing to nest them in a tree structure. You can assign the submodules as regular attributes:

import torch.nn as nn
import torch.nn.functional as F

class Model(nn.Module):
    def __init__(self):
        super().__init__()
        self.conv1 = nn.Conv2d(1, 20, 5)
        self.conv2 = nn.Conv2d(20, 20, 5)

    def forward(self, x):
        x = F.relu(self.conv1(x))
        return F.relu(self.conv2(x))

Submodules assigned in this way will be registered, and will have their parameters converted too when you call to(), etc.

Note

As per the example above, an __init__() call to the parent class must be made before assignment on the child.

Variables:

training (bool) – Boolean represents whether this module is in training or evaluation mode.

abstract property label_requirement: List[deepmd.utils.data.DataRequirementItem]

Return data label requirements needed for this loss calculation.

abstract forward(input_dict, model, label, natoms, learning_rate)

Return loss .

static display_if_exist(loss: torch.Tensor, find_property: float) → torch.Tensor

Display NaN if labeled property is not found.

Parameters:

losstorch.Tensor

the loss tensor

find_propertyfloat

whether the property is found

class deepmd.pt.loss.TensorLoss(tensor_name: str, tensor_size: int, label_name: str, pref_atomic: float = 0.0, pref: float = 0.0, inference=False, **kwargs)

Bases: deepmd.pt.loss.loss.TaskLoss

Base class for all neural network modules.

Your models should also subclass this class.

Modules can also contain other Modules, allowing to nest them in a tree structure. You can assign the submodules as regular attributes:

import torch.nn as nn
import torch.nn.functional as F

class Model(nn.Module):
    def __init__(self):
        super().__init__()
        self.conv1 = nn.Conv2d(1, 20, 5)
        self.conv2 = nn.Conv2d(20, 20, 5)

    def forward(self, x):
        x = F.relu(self.conv1(x))
        return F.relu(self.conv2(x))

Submodules assigned in this way will be registered, and will have their parameters converted too when you call to(), etc.

Note

As per the example above, an __init__() call to the parent class must be made before assignment on the child.

Variables:

training (bool) – Boolean represents whether this module is in training or evaluation mode.

property label_requirement: List[deepmd.utils.data.DataRequirementItem]

Return data label requirements needed for this loss calculation.

forward(input_dict, model, label, natoms, learning_rate=0.0, mae=False)

Return loss on local and global tensors.

Parameters:

input_dictdict[str, torch.Tensor]

Model inputs.

modeltorch.nn.Module

Model to be used to output the predictions.

labeldict[str, torch.Tensor]

Labels.

natomsint

The local atom number.

Returns:

model_pred: dict[str, torch.Tensor]

Model predictions.

loss: torch.Tensor

Loss for model to minimize.

more_loss: dict[str, torch.Tensor]

Other losses for display.