cait.models¶

class cait.models.LSTMModule(input_size, hidden_size, num_layers, seq_steps, nmbr_out, label_keys, feature_keys, lr, device_name='cpu', is_classifier=True, down=1, down_keys=None, norm_vals=None, offset_keys=None, weight_decay=1e-05, bidirectional=False, norm_type='minmax', lr_scheduler=True, indiv_norm=False, attention=False)[source]¶

Bases: pytorch_lightning.core.lightning.LightningModule

Lightning module for the training of an LSTM model for classification or regression. For classification, the classes need to get one hot encoded, best with the corresponding transform.

Parameters

input_size (int) – The number of features that get passed to the LSTM in one time step.
hidden_size (int) – The number of nodes in the hidden layer of the lstm.
num_layers (int) – The number of LSTM layers.
seq_steps (int) – The number of time steps.
device_name (string, either 'cpu' or 'cude') – The device on that the NN is trained.
nmbr_out (int) – The number of output nodes the last linear layer after the lstm has.
label_keys (list of strings) – The keys of the dataset that are used as labels.
feature_keys (list of strings) – The keys of the dataset that are used as nn inputs.
lr (float between 0 and 1) – The learning rate for the neural network training.
is_classifier (bool) – If true, the output of the nn gets an additional softmax activation.
down (int) – The downsample factor of the training data set, if one is applied.
down_keys (list of string) – The keys of the data that is to downsample (usually the event time series).
norm_vals (dictionary, every enty is a list of 2 ints (mean, std)) – The keys of this dictionary get scaled in the sample with (x - mu)/sigma.
offset_keys (list of strings) – The keys in the sample from that we want to subtract the baseline offset level.
weight_decay (float) – The weight decay parameter for the optimizer.
bidirectional (bool) – If true, a bidirectional LSTM is used.
norm_type (string) – Either ‘z’ (mu=0, sigma=1) or ‘minmax’ (min=0, max=1). The type of normalization.
lr_scheduler (bool) – If true, a learning rate scheduler is used.
indiv_norm (bool) – If true, every event is divide by its maximal value before passing into the network.
attention (bool) – If activated, an attention layer is added before passing into the model.

configure_optimizers(lr=None, weight_decay=None)[source]¶

Choose what optimizers and learning-rate schedulers to use in your optimization. Normally you’d need one. But in the case of GANs or similar you might have multiple.

Returns

Any of these 6 options.

Single optimizer.
List or Tuple of optimizers.
Two lists - The first list has multiple optimizers, and the second has multiple LR schedulers (or multiple lr_dict).
Dictionary, with an "optimizer" key, and (optionally) a "lr_scheduler" key whose value is a single LR scheduler or lr_dict.
Tuple of dictionaries as described above, with an optional "frequency" key.
None - Fit will run without any optimizer.

Note

The lr_dict is a dictionary which contains the scheduler and its associated configuration. The default configuration is shown below.

lr_dict = {
    'scheduler': lr_scheduler, # The LR scheduler instance (required)
    # The unit of the scheduler's step size, could also be 'step'
    'interval': 'epoch',
    'frequency': 1, # The frequency of the scheduler
    'monitor': 'val_loss', # Metric for `ReduceLROnPlateau` to monitor
    'strict': True, # Whether to crash the training if `monitor` is not found
    'name': None, # Custom name for `LearningRateMonitor` to use
}

Only the "scheduler" key is required, the rest will be set to the defaults above.

Note

The frequency value specified in a dict along with the optimizer key is an int corresponding to the number of sequential batches optimized with the specific optimizer. It should be given to none or to all of the optimizers. There is a difference between passing multiple optimizers in a list, and passing multiple optimizers in dictionaries with a frequency of 1: In the former case, all optimizers will operate on the given batch in each optimization step. In the latter, only one optimizer will operate on the given batch at every step. This is different from the frequency value specified in the lr_dict mentioned below.

def configure_optimizers(self):
    optimizer_one = torch.optim.SGD(self.model.parameters(), lr=0.01)
    optimizer_two = torch.optim.SGD(self.model.parameters(), lr=0.01)
    return [
        {'optimizer': optimizer_one, 'frequency': 5},
        {'optimizer': optimizer_two, 'frequency': 10},
    ]

In this example, the first optimizer will be used for the first 5 steps, the second optimizer for the next 10 steps and that cycle will continue. If an LR scheduler is specified for an optimizer using the lr_scheduler key in the above dict, the scheduler will only be updated when its optimizer is being used.

Examples:

# most cases
def configure_optimizers(self):
    return Adam(self.parameters(), lr=1e-3)

# multiple optimizer case (e.g.: GAN)
def configure_optimizers(self):
    gen_opt = Adam(self.model_gen.parameters(), lr=0.01)
    dis_opt = Adam(self.model_dis.parameters(), lr=0.02)
    return gen_opt, dis_opt

# example with learning rate schedulers
def configure_optimizers(self):
    gen_opt = Adam(self.model_gen.parameters(), lr=0.01)
    dis_opt = Adam(self.model_dis.parameters(), lr=0.02)
    dis_sch = CosineAnnealing(dis_opt, T_max=10)
    return [gen_opt, dis_opt], [dis_sch]

# example with step-based learning rate schedulers
def configure_optimizers(self):
    gen_opt = Adam(self.model_gen.parameters(), lr=0.01)
    dis_opt = Adam(self.model_dis.parameters(), lr=0.02)
    gen_sch = {'scheduler': ExponentialLR(gen_opt, 0.99),
               'interval': 'step'}  # called after each training step
    dis_sch = CosineAnnealing(dis_opt, T_max=10) # called every epoch
    return [gen_opt, dis_opt], [gen_sch, dis_sch]

# example with optimizer frequencies
# see training procedure in `Improved Training of Wasserstein GANs`, Algorithm 1
# https://arxiv.org/abs/1704.00028
def configure_optimizers(self):
    gen_opt = Adam(self.model_gen.parameters(), lr=0.01)
    dis_opt = Adam(self.model_dis.parameters(), lr=0.02)
    n_critic = 5
    return (
        {'optimizer': dis_opt, 'frequency': n_critic},
        {'optimizer': gen_opt, 'frequency': 1}
    )

Note

Some things to know:

Lightning calls .backward() and .step() on each optimizer and learning rate scheduler as needed.
If you use 16-bit precision (precision=16), Lightning will automatically handle the optimizers.
If you use multiple optimizers, training_step() will have an additional optimizer_idx parameter.
If you use torch.optim.LBFGS, Lightning handles the closure function automatically for you.
If you use multiple optimizers, gradients will be calculated only for the parameters of current optimizer at each training step.
If you need to control how often those optimizers step or override the default .step() schedule, override the optimizer_step() hook.

forward(x)[source]¶

The forward pass in the neural network.

Parameters: x (torch tensor of size (batchsize, nmbr_features)) – The input features.
Returns: The ouput of the neural network.
Return type: torch tensor of size (batchsize, nmbr_outputs)

loss_function(logits, labels)[source]¶

predict(sample)[source]¶

Give a prediction for incoming data array or batch of arrays, does all essential transforms.

Parameters: sample (1D numpy array or batch of arrays, i.e. then 2D array) – The features for one (1D case) or more (2D case) samples.
Returns: The prediction.
Return type: torch tensor of size (batchsize - 1 if no batch, nn_output_size)

test_step(batch, batch_idx)[source]¶

Operates on a single batch of data from the test set. In this step you’d normally generate examples or calculate anything of interest such as accuracy.

# the pseudocode for these calls
test_outs = []
for test_batch in test_data:
    out = test_step(test_batch)
    test_outs.append(out)
test_epoch_end(test_outs)

Parameters

batch (Tensor | (Tensor, …) | [Tensor, …]) – The output of your DataLoader. A tensor, tuple or list.
batch_idx (int) – The index of this batch.
dataloader_idx (int) – The index of the dataloader that produced this batch (only if multiple test dataloaders used).

Returns

Any of.

Any object or value

None - Testing will skip to the next batch

# if you have one test dataloader:
def test_step(self, batch, batch_idx)

# if you have multiple test dataloaders:
def test_step(self, batch, batch_idx, dataloader_idx)

Examples:

# CASE 1: A single test dataset
def test_step(self, batch, batch_idx):
    x, y = batch

    # implement your own
    out = self(x)
    loss = self.loss(out, y)

    # log 6 example images
    # or generated text... or whatever
    sample_imgs = x[:6]
    grid = torchvision.utils.make_grid(sample_imgs)
    self.logger.experiment.add_image('example_images', grid, 0)

    # calculate acc
    labels_hat = torch.argmax(out, dim=1)
    test_acc = torch.sum(y == labels_hat).item() / (len(y) * 1.0)

    # log the outputs!
    self.log_dict({'test_loss': loss, 'test_acc': test_acc})

If you pass in multiple test dataloaders, test_step() will have an additional argument.

# CASE 2: multiple test dataloaders
def test_step(self, batch, batch_idx, dataloader_idx):
    # dataloader_idx tells you which dataset this is.

Note

If you don’t need to test you don’t need to implement this method.

Note

When the test_step() is called, the model has been put in eval mode and PyTorch gradients have been disabled. At the end of the test epoch, the model goes back to training mode and gradients are enabled.

training: bool¶

training_step(batch, batch_idx)[source]¶

Here you compute and return the training loss and some additional metrics for e.g. the progress bar or logger.

Parameters

batch (Tensor | (Tensor, …) | [Tensor, …]) – The output of your DataLoader. A tensor, tuple or list.
batch_idx (int) – Integer displaying index of this batch
optimizer_idx (int) – When using multiple optimizers, this argument will also be present.
hiddens (Tensor) – Passed in if :paramref:`~pytorch_lightning.core.lightning.LightningModule.truncated_bptt_steps` > 0.

Returns

Any of.

Tensor - The loss tensor
dict - A dictionary. Can include any keys, but must include the key 'loss'
None - Training will skip to the next batch

Note

Returning None is currently not supported for multi-GPU or TPU, or with 16-bit precision enabled.

In this step you’d normally do the forward pass and calculate the loss for a batch. You can also do fancier things like multiple forward passes or something model specific.

Example:

def training_step(self, batch, batch_idx):
    x, y, z = batch
    out = self.encoder(x)
    loss = self.loss(out, x)
    return loss

If you define multiple optimizers, this step will be called with an additional optimizer_idx parameter.

# Multiple optimizers (e.g.: GANs)
def training_step(self, batch, batch_idx, optimizer_idx):
    if optimizer_idx == 0:
        # do training_step with encoder
    if optimizer_idx == 1:
        # do training_step with decoder

If you add truncated back propagation through time you will also get an additional argument with the hidden states of the previous step.

# Truncated back-propagation through time
def training_step(self, batch, batch_idx, hiddens):
    # hiddens are the hidden states from the previous truncated backprop step
    ...
    out, hiddens = self.lstm(data, hiddens)
    ...
    return {'loss': loss, 'hiddens': hiddens}

Note

The loss value shown in the progress bar is smoothed (averaged) over the last values, so it differs from the actual loss returned in train/validation step.

validation_step(val_batch, batch_idx)[source]¶

Operates on a single batch of data from the validation set. In this step you’d might generate examples or calculate anything of interest like accuracy.

# the pseudocode for these calls
val_outs = []
for val_batch in val_data:
    out = validation_step(val_batch)
    val_outs.append(out)
validation_epoch_end(val_outs)

Parameters

batch (Tensor | (Tensor, …) | [Tensor, …]) – The output of your DataLoader. A tensor, tuple or list.
batch_idx (int) – The index of this batch
dataloader_idx (int) – The index of the dataloader that produced this batch (only if multiple val dataloaders used)

Returns

Any of.

Any object or value

None - Validation will skip to the next batch

# pseudocode of order
val_outs = []
for val_batch in val_data:
    out = validation_step(val_batch)
    if defined('validation_step_end'):
        out = validation_step_end(out)
    val_outs.append(out)
val_outs = validation_epoch_end(val_outs)

# if you have one val dataloader:
def validation_step(self, batch, batch_idx)

# if you have multiple val dataloaders:
def validation_step(self, batch, batch_idx, dataloader_idx)

Examples:

# CASE 1: A single validation dataset
def validation_step(self, batch, batch_idx):
    x, y = batch

    # implement your own
    out = self(x)
    loss = self.loss(out, y)

    # log 6 example images
    # or generated text... or whatever
    sample_imgs = x[:6]
    grid = torchvision.utils.make_grid(sample_imgs)
    self.logger.experiment.add_image('example_images', grid, 0)

    # calculate acc
    labels_hat = torch.argmax(out, dim=1)
    val_acc = torch.sum(y == labels_hat).item() / (len(y) * 1.0)

    # log the outputs!
    self.log_dict({'val_loss': loss, 'val_acc': val_acc})

If you pass in multiple val dataloaders, validation_step() will have an additional argument.

# CASE 2: multiple validation dataloaders
def validation_step(self, batch, batch_idx, dataloader_idx):
    # dataloader_idx tells you which dataset this is.

Note

If you don’t need to validate you don’t need to implement this method.

Note

When the validation_step() is called, the model has been put in eval mode and PyTorch gradients have been disabled. At the end of validation, the model goes back to training mode and gradients are enabled.

class cait.models.RNNModule(input_size, hidden_size, num_layers, seq_steps, nmbr_out, label_keys, feature_keys, lr, device_name='cpu', is_classifier=True, down=1, down_keys=None, norm_vals=None, offset_keys=None, weight_decay=1e-05)[source]¶

Bases: pytorch_lightning.core.lightning.LightningModule

Lightning module for the training of an RNN model for classification or regression.

For classification, the classes need to get one hot encoded, best with the corresponding transform.

Parameters

input_size (int) – The number of features that get passed to the RNN in one time step.
hidden_size (int) – The number of nodes in the hidden layer of the RNN.
num_layers (int) – The number of RNN layers.
seq_steps (int) – The number of time steps.
device_name (string, either 'cpu' or 'cude') – The device on that the NN is trained.
nmbr_out (int) – The number of output nodes the last linear layer after the RNN has.
label_keys (list of strings) – The keys of the dataset that are used as labels.
feature_keys (list of strings) – The keys of the dataset that are used as nn inputs.
lr (float between 0 and 1) – The learning rate for the neural network training.
is_classifier (bool) – If true, the output of the nn gets an additional softmax activation.
down (int) – The downsample factor of the training data set, if one is applied.
down_keys (list of string) – The keys of the data that is to downsample (usually the event time series).
norm_vals (dictionary, every enty is a list of 2 ints (mean, std)) – The keys of this dictionary get scaled in the sample with (x - mu)/sigma.
offset_keys (list of strings) – The keys in the sample from that we want to subtract the baseline offset level.
weight_decay (float) – The weight decay parameter for the optimizer.

configure_optimizers(lr=None, weight_decay=None)[source]¶

Choose what optimizers and learning-rate schedulers to use in your optimization. Normally you’d need one. But in the case of GANs or similar you might have multiple.

Returns

Any of these 6 options.

Single optimizer.
List or Tuple of optimizers.
Two lists - The first list has multiple optimizers, and the second has multiple LR schedulers (or multiple lr_dict).
Dictionary, with an "optimizer" key, and (optionally) a "lr_scheduler" key whose value is a single LR scheduler or lr_dict.
Tuple of dictionaries as described above, with an optional "frequency" key.
None - Fit will run without any optimizer.

Note

The lr_dict is a dictionary which contains the scheduler and its associated configuration. The default configuration is shown below.

lr_dict = {
    'scheduler': lr_scheduler, # The LR scheduler instance (required)
    # The unit of the scheduler's step size, could also be 'step'
    'interval': 'epoch',
    'frequency': 1, # The frequency of the scheduler
    'monitor': 'val_loss', # Metric for `ReduceLROnPlateau` to monitor
    'strict': True, # Whether to crash the training if `monitor` is not found
    'name': None, # Custom name for `LearningRateMonitor` to use
}

Only the "scheduler" key is required, the rest will be set to the defaults above.

Note

The frequency value specified in a dict along with the optimizer key is an int corresponding to the number of sequential batches optimized with the specific optimizer. It should be given to none or to all of the optimizers. There is a difference between passing multiple optimizers in a list, and passing multiple optimizers in dictionaries with a frequency of 1: In the former case, all optimizers will operate on the given batch in each optimization step. In the latter, only one optimizer will operate on the given batch at every step. This is different from the frequency value specified in the lr_dict mentioned below.

def configure_optimizers(self):
    optimizer_one = torch.optim.SGD(self.model.parameters(), lr=0.01)
    optimizer_two = torch.optim.SGD(self.model.parameters(), lr=0.01)
    return [
        {'optimizer': optimizer_one, 'frequency': 5},
        {'optimizer': optimizer_two, 'frequency': 10},
    ]

In this example, the first optimizer will be used for the first 5 steps, the second optimizer for the next 10 steps and that cycle will continue. If an LR scheduler is specified for an optimizer using the lr_scheduler key in the above dict, the scheduler will only be updated when its optimizer is being used.

Examples:

# most cases
def configure_optimizers(self):
    return Adam(self.parameters(), lr=1e-3)

# multiple optimizer case (e.g.: GAN)
def configure_optimizers(self):
    gen_opt = Adam(self.model_gen.parameters(), lr=0.01)
    dis_opt = Adam(self.model_dis.parameters(), lr=0.02)
    return gen_opt, dis_opt

# example with learning rate schedulers
def configure_optimizers(self):
    gen_opt = Adam(self.model_gen.parameters(), lr=0.01)
    dis_opt = Adam(self.model_dis.parameters(), lr=0.02)
    dis_sch = CosineAnnealing(dis_opt, T_max=10)
    return [gen_opt, dis_opt], [dis_sch]

# example with step-based learning rate schedulers
def configure_optimizers(self):
    gen_opt = Adam(self.model_gen.parameters(), lr=0.01)
    dis_opt = Adam(self.model_dis.parameters(), lr=0.02)
    gen_sch = {'scheduler': ExponentialLR(gen_opt, 0.99),
               'interval': 'step'}  # called after each training step
    dis_sch = CosineAnnealing(dis_opt, T_max=10) # called every epoch
    return [gen_opt, dis_opt], [gen_sch, dis_sch]

# example with optimizer frequencies
# see training procedure in `Improved Training of Wasserstein GANs`, Algorithm 1
# https://arxiv.org/abs/1704.00028
def configure_optimizers(self):
    gen_opt = Adam(self.model_gen.parameters(), lr=0.01)
    dis_opt = Adam(self.model_dis.parameters(), lr=0.02)
    n_critic = 5
    return (
        {'optimizer': dis_opt, 'frequency': n_critic},
        {'optimizer': gen_opt, 'frequency': 1}
    )

Note

Some things to know:

Lightning calls .backward() and .step() on each optimizer and learning rate scheduler as needed.
If you use 16-bit precision (precision=16), Lightning will automatically handle the optimizers.
If you use multiple optimizers, training_step() will have an additional optimizer_idx parameter.
If you use torch.optim.LBFGS, Lightning handles the closure function automatically for you.
If you use multiple optimizers, gradients will be calculated only for the parameters of current optimizer at each training step.
If you need to control how often those optimizers step or override the default .step() schedule, override the optimizer_step() hook.

forward(x)[source]¶

The forward pass in the neural network.

Parameters: x (torch tensor of size (batchsize, nmbr_features)) – The input features.
Returns: The ouput of the neural network.
Return type: torch tensor of size (batchsize, nmbr_outputs)

loss_function(logits, labels)[source]¶

predict(sample)[source]¶

Give a prediction for incoming data array or batch of arrays, does all essential transforms

Parameters: sample (1D numpy array or batch of arrays, i.e. then 2D array) – The features for one (1D case) or more (2D case) samples.
Returns: The prediction.
Return type: torch tensor of size (batchsize - 1 if no batch, nn_output_size)

test_step(batch, batch_idx)[source]¶

Operates on a single batch of data from the test set. In this step you’d normally generate examples or calculate anything of interest such as accuracy.

# the pseudocode for these calls
test_outs = []
for test_batch in test_data:
    out = test_step(test_batch)
    test_outs.append(out)
test_epoch_end(test_outs)

Parameters

batch (Tensor | (Tensor, …) | [Tensor, …]) – The output of your DataLoader. A tensor, tuple or list.
batch_idx (int) – The index of this batch.
dataloader_idx (int) – The index of the dataloader that produced this batch (only if multiple test dataloaders used).

Returns

Any of.

Any object or value

None - Testing will skip to the next batch

# if you have one test dataloader:
def test_step(self, batch, batch_idx)

# if you have multiple test dataloaders:
def test_step(self, batch, batch_idx, dataloader_idx)

Examples:

# CASE 1: A single test dataset
def test_step(self, batch, batch_idx):
    x, y = batch

    # implement your own
    out = self(x)
    loss = self.loss(out, y)

    # log 6 example images
    # or generated text... or whatever
    sample_imgs = x[:6]
    grid = torchvision.utils.make_grid(sample_imgs)
    self.logger.experiment.add_image('example_images', grid, 0)

    # calculate acc
    labels_hat = torch.argmax(out, dim=1)
    test_acc = torch.sum(y == labels_hat).item() / (len(y) * 1.0)

    # log the outputs!
    self.log_dict({'test_loss': loss, 'test_acc': test_acc})

If you pass in multiple test dataloaders, test_step() will have an additional argument.

# CASE 2: multiple test dataloaders
def test_step(self, batch, batch_idx, dataloader_idx):
    # dataloader_idx tells you which dataset this is.

Note

If you don’t need to test you don’t need to implement this method.

Note

When the test_step() is called, the model has been put in eval mode and PyTorch gradients have been disabled. At the end of the test epoch, the model goes back to training mode and gradients are enabled.

training: bool¶

training_step(batch, batch_idx)[source]¶

Here you compute and return the training loss and some additional metrics for e.g. the progress bar or logger.

Parameters

batch (Tensor | (Tensor, …) | [Tensor, …]) – The output of your DataLoader. A tensor, tuple or list.
batch_idx (int) – Integer displaying index of this batch
optimizer_idx (int) – When using multiple optimizers, this argument will also be present.
hiddens (Tensor) – Passed in if :paramref:`~pytorch_lightning.core.lightning.LightningModule.truncated_bptt_steps` > 0.

Returns

Any of.

Tensor - The loss tensor
dict - A dictionary. Can include any keys, but must include the key 'loss'
None - Training will skip to the next batch

Note

Returning None is currently not supported for multi-GPU or TPU, or with 16-bit precision enabled.

In this step you’d normally do the forward pass and calculate the loss for a batch. You can also do fancier things like multiple forward passes or something model specific.

Example:

def training_step(self, batch, batch_idx):
    x, y, z = batch
    out = self.encoder(x)
    loss = self.loss(out, x)
    return loss

If you define multiple optimizers, this step will be called with an additional optimizer_idx parameter.

# Multiple optimizers (e.g.: GANs)
def training_step(self, batch, batch_idx, optimizer_idx):
    if optimizer_idx == 0:
        # do training_step with encoder
    if optimizer_idx == 1:
        # do training_step with decoder

If you add truncated back propagation through time you will also get an additional argument with the hidden states of the previous step.

# Truncated back-propagation through time
def training_step(self, batch, batch_idx, hiddens):
    # hiddens are the hidden states from the previous truncated backprop step
    ...
    out, hiddens = self.lstm(data, hiddens)
    ...
    return {'loss': loss, 'hiddens': hiddens}

Note

The loss value shown in the progress bar is smoothed (averaged) over the last values, so it differs from the actual loss returned in train/validation step.

validation_step(val_batch, batch_idx)[source]¶

Operates on a single batch of data from the validation set. In this step you’d might generate examples or calculate anything of interest like accuracy.

# the pseudocode for these calls
val_outs = []
for val_batch in val_data:
    out = validation_step(val_batch)
    val_outs.append(out)
validation_epoch_end(val_outs)

Parameters

batch (Tensor | (Tensor, …) | [Tensor, …]) – The output of your DataLoader. A tensor, tuple or list.
batch_idx (int) – The index of this batch
dataloader_idx (int) – The index of the dataloader that produced this batch (only if multiple val dataloaders used)

Returns

Any of.

Any object or value

None - Validation will skip to the next batch

# pseudocode of order
val_outs = []
for val_batch in val_data:
    out = validation_step(val_batch)
    if defined('validation_step_end'):
        out = validation_step_end(out)
    val_outs.append(out)
val_outs = validation_epoch_end(val_outs)

# if you have one val dataloader:
def validation_step(self, batch, batch_idx)

# if you have multiple val dataloaders:
def validation_step(self, batch, batch_idx, dataloader_idx)

Examples:

# CASE 1: A single validation dataset
def validation_step(self, batch, batch_idx):
    x, y = batch

    # implement your own
    out = self(x)
    loss = self.loss(out, y)

    # log 6 example images
    # or generated text... or whatever
    sample_imgs = x[:6]
    grid = torchvision.utils.make_grid(sample_imgs)
    self.logger.experiment.add_image('example_images', grid, 0)

    # calculate acc
    labels_hat = torch.argmax(out, dim=1)
    val_acc = torch.sum(y == labels_hat).item() / (len(y) * 1.0)

    # log the outputs!
    self.log_dict({'val_loss': loss, 'val_acc': val_acc})

If you pass in multiple val dataloaders, validation_step() will have an additional argument.

# CASE 2: multiple validation dataloaders
def validation_step(self, batch, batch_idx, dataloader_idx):
    # dataloader_idx tells you which dataset this is.

Note

If you don’t need to validate you don’t need to implement this method.

Note

When the validation_step() is called, the model has been put in eval mode and PyTorch gradients have been disabled. At the end of validation, the model goes back to training mode and gradients are enabled.

class cait.models.TransformerModule(input_size, d_model, number_heads, dim_feedforward, num_layers, nmbr_out, seq_steps, device_name, label_keys, feature_keys, lr, is_classifier, down, down_keys, offset_keys, norm_vals, weight_decay=1e-05, dropout=0.5, norm_type='minmax', pos_enc=True, lr_scheduler=True)[source]¶

Bases: pytorch_lightning.core.lightning.LightningModule

Lightning module for the training of an Transformer Encoder model for classification or regression. For classification, the classes need to get one hot encoded, best with the corresponding transform.

Parameters

input_size (int) – The number of features that get passed to the Model in one time step.
d_model (int) – The dimensions of the model.
number_heads (int) – The number of heads for the attention layer.
dim_feedforward (int) – The dimensions in the feed forward net.
hidden_size (int) – The number of nodes in the hidden layer of the lstm.
num_layers (int) – The number of LSTM layers.
seq_steps (int) – The number of time steps.
device_name (string) – The device on that the NN is trained. Either ‘cpu’ or ‘cuda’.
nmbr_out (int) – The number of output nodes the last linear layer after the lstm has.
label_keys (list of strings) – The keys of the dataset that are used as labels.
feature_keys (list of strings) – The keys of the dataset that are used as nn inputs.
lr (float between 0 and 1) – The learning rate for the neural network training.
is_classifier (bool) – If true, the output of the nn gets an additional softmax activation.
down (int) – The downsample factor of the training data set, if one is applied.
down_keys (list of string) – The keys of the data that is to downsample (usually the event time series).
norm_vals (dictionary, every enty is a list of 2 ints (mean, std)) – The keys of this dictionary get scaled in the sample with (x - mu)/sigma.
offset_keys (list of strings) – The keys in the sample from that we want to subtract the baseline offset level.
weight_decay (float) – The weight decay parameter for the optimizer.
dropout (float) – The share of weights that is set to zero in the dropout layer.
norm_type (string) – Either ‘z’ (mu=0, sigma=1) or ‘minmax’ (min=0, max=1). The type of normalization.
pos_enc (bool) – If true, we include a positional encoding layer.
lr_scheduler (bool) – If true, a learning rate scheduler is used.

configure_optimizers(lr=None, weight_decay=None)[source]¶

Choose what optimizers and learning-rate schedulers to use in your optimization. Normally you’d need one. But in the case of GANs or similar you might have multiple.

Returns

Any of these 6 options.

Single optimizer.
List or Tuple of optimizers.
Two lists - The first list has multiple optimizers, and the second has multiple LR schedulers (or multiple lr_dict).
Dictionary, with an "optimizer" key, and (optionally) a "lr_scheduler" key whose value is a single LR scheduler or lr_dict.
Tuple of dictionaries as described above, with an optional "frequency" key.
None - Fit will run without any optimizer.

Note

The lr_dict is a dictionary which contains the scheduler and its associated configuration. The default configuration is shown below.

lr_dict = {
    'scheduler': lr_scheduler, # The LR scheduler instance (required)
    # The unit of the scheduler's step size, could also be 'step'
    'interval': 'epoch',
    'frequency': 1, # The frequency of the scheduler
    'monitor': 'val_loss', # Metric for `ReduceLROnPlateau` to monitor
    'strict': True, # Whether to crash the training if `monitor` is not found
    'name': None, # Custom name for `LearningRateMonitor` to use
}

Only the "scheduler" key is required, the rest will be set to the defaults above.

Note

The frequency value specified in a dict along with the optimizer key is an int corresponding to the number of sequential batches optimized with the specific optimizer. It should be given to none or to all of the optimizers. There is a difference between passing multiple optimizers in a list, and passing multiple optimizers in dictionaries with a frequency of 1: In the former case, all optimizers will operate on the given batch in each optimization step. In the latter, only one optimizer will operate on the given batch at every step. This is different from the frequency value specified in the lr_dict mentioned below.

def configure_optimizers(self):
    optimizer_one = torch.optim.SGD(self.model.parameters(), lr=0.01)
    optimizer_two = torch.optim.SGD(self.model.parameters(), lr=0.01)
    return [
        {'optimizer': optimizer_one, 'frequency': 5},
        {'optimizer': optimizer_two, 'frequency': 10},
    ]

In this example, the first optimizer will be used for the first 5 steps, the second optimizer for the next 10 steps and that cycle will continue. If an LR scheduler is specified for an optimizer using the lr_scheduler key in the above dict, the scheduler will only be updated when its optimizer is being used.

Examples:

# most cases
def configure_optimizers(self):
    return Adam(self.parameters(), lr=1e-3)

# multiple optimizer case (e.g.: GAN)
def configure_optimizers(self):
    gen_opt = Adam(self.model_gen.parameters(), lr=0.01)
    dis_opt = Adam(self.model_dis.parameters(), lr=0.02)
    return gen_opt, dis_opt

# example with learning rate schedulers
def configure_optimizers(self):
    gen_opt = Adam(self.model_gen.parameters(), lr=0.01)
    dis_opt = Adam(self.model_dis.parameters(), lr=0.02)
    dis_sch = CosineAnnealing(dis_opt, T_max=10)
    return [gen_opt, dis_opt], [dis_sch]

# example with step-based learning rate schedulers
def configure_optimizers(self):
    gen_opt = Adam(self.model_gen.parameters(), lr=0.01)
    dis_opt = Adam(self.model_dis.parameters(), lr=0.02)
    gen_sch = {'scheduler': ExponentialLR(gen_opt, 0.99),
               'interval': 'step'}  # called after each training step
    dis_sch = CosineAnnealing(dis_opt, T_max=10) # called every epoch
    return [gen_opt, dis_opt], [gen_sch, dis_sch]

# example with optimizer frequencies
# see training procedure in `Improved Training of Wasserstein GANs`, Algorithm 1
# https://arxiv.org/abs/1704.00028
def configure_optimizers(self):
    gen_opt = Adam(self.model_gen.parameters(), lr=0.01)
    dis_opt = Adam(self.model_dis.parameters(), lr=0.02)
    n_critic = 5
    return (
        {'optimizer': dis_opt, 'frequency': n_critic},
        {'optimizer': gen_opt, 'frequency': 1}
    )

Note

Some things to know:

Lightning calls .backward() and .step() on each optimizer and learning rate scheduler as needed.
If you use 16-bit precision (precision=16), Lightning will automatically handle the optimizers.
If you use multiple optimizers, training_step() will have an additional optimizer_idx parameter.
If you use torch.optim.LBFGS, Lightning handles the closure function automatically for you.
If you use multiple optimizers, gradients will be calculated only for the parameters of current optimizer at each training step.
If you need to control how often those optimizers step or override the default .step() schedule, override the optimizer_step() hook.

forward(src, src_mask=None)[source]¶

The forward pass in the neural network

Parameters: x (torch tensor of size (batchsize, nmbr_features)) – the input features
Returns: the ouput of the neural network
Return type: torch tensor of size (batchsize, nmbr_outputs)

generate_square_subsequent_mask(sz)[source]¶

loss_function(logits, labels)[source]¶

Calculates the loss value, for classfiers the negative log likelihood, for regressors the MSE.

Parameters

logits (float) – The output values of the neural network.
labels (float) – The labels, e.g. the objective values or classes.

Returns

The loss value

Return type

float

predict(sample)[source]¶

Give a prediction for incoming data array or batch of arrays, does all essential transforms

Parameters: sample (1D numpy array or batch of arrays, i.e. then 2D array) – the features for one (1D case) or more (2D case) samples
Returns: the prediction
Return type: torch tensor of size (batchsize - 1 if no batch, nn_output_size)

test_step(batch, batch_idx)[source]¶

Operates on a single batch of data from the test set. In this step you’d normally generate examples or calculate anything of interest such as accuracy.

# the pseudocode for these calls
test_outs = []
for test_batch in test_data:
    out = test_step(test_batch)
    test_outs.append(out)
test_epoch_end(test_outs)

Parameters

batch (Tensor | (Tensor, …) | [Tensor, …]) – The output of your DataLoader. A tensor, tuple or list.
batch_idx (int) – The index of this batch.
dataloader_idx (int) – The index of the dataloader that produced this batch (only if multiple test dataloaders used).

Returns

Any of.

Any object or value

None - Testing will skip to the next batch

# if you have one test dataloader:
def test_step(self, batch, batch_idx)

# if you have multiple test dataloaders:
def test_step(self, batch, batch_idx, dataloader_idx)

Examples:

# CASE 1: A single test dataset
def test_step(self, batch, batch_idx):
    x, y = batch

    # implement your own
    out = self(x)
    loss = self.loss(out, y)

    # log 6 example images
    # or generated text... or whatever
    sample_imgs = x[:6]
    grid = torchvision.utils.make_grid(sample_imgs)
    self.logger.experiment.add_image('example_images', grid, 0)

    # calculate acc
    labels_hat = torch.argmax(out, dim=1)
    test_acc = torch.sum(y == labels_hat).item() / (len(y) * 1.0)

    # log the outputs!
    self.log_dict({'test_loss': loss, 'test_acc': test_acc})

If you pass in multiple test dataloaders, test_step() will have an additional argument.

# CASE 2: multiple test dataloaders
def test_step(self, batch, batch_idx, dataloader_idx):
    # dataloader_idx tells you which dataset this is.

Note

If you don’t need to test you don’t need to implement this method.

Note

When the test_step() is called, the model has been put in eval mode and PyTorch gradients have been disabled. At the end of the test epoch, the model goes back to training mode and gradients are enabled.

training: bool¶

training_step(batch, batch_idx)[source]¶

Here you compute and return the training loss and some additional metrics for e.g. the progress bar or logger.

Parameters

batch (Tensor | (Tensor, …) | [Tensor, …]) – The output of your DataLoader. A tensor, tuple or list.
batch_idx (int) – Integer displaying index of this batch
optimizer_idx (int) – When using multiple optimizers, this argument will also be present.
hiddens (Tensor) – Passed in if :paramref:`~pytorch_lightning.core.lightning.LightningModule.truncated_bptt_steps` > 0.

Returns

Any of.

Tensor - The loss tensor
dict - A dictionary. Can include any keys, but must include the key 'loss'
None - Training will skip to the next batch

Note

Returning None is currently not supported for multi-GPU or TPU, or with 16-bit precision enabled.

In this step you’d normally do the forward pass and calculate the loss for a batch. You can also do fancier things like multiple forward passes or something model specific.

Example:

def training_step(self, batch, batch_idx):
    x, y, z = batch
    out = self.encoder(x)
    loss = self.loss(out, x)
    return loss

If you define multiple optimizers, this step will be called with an additional optimizer_idx parameter.

# Multiple optimizers (e.g.: GANs)
def training_step(self, batch, batch_idx, optimizer_idx):
    if optimizer_idx == 0:
        # do training_step with encoder
    if optimizer_idx == 1:
        # do training_step with decoder

If you add truncated back propagation through time you will also get an additional argument with the hidden states of the previous step.

# Truncated back-propagation through time
def training_step(self, batch, batch_idx, hiddens):
    # hiddens are the hidden states from the previous truncated backprop step
    ...
    out, hiddens = self.lstm(data, hiddens)
    ...
    return {'loss': loss, 'hiddens': hiddens}

Note

The loss value shown in the progress bar is smoothed (averaged) over the last values, so it differs from the actual loss returned in train/validation step.

validation_step(val_batch, batch_idx)[source]¶

Operates on a single batch of data from the validation set. In this step you’d might generate examples or calculate anything of interest like accuracy.

# the pseudocode for these calls
val_outs = []
for val_batch in val_data:
    out = validation_step(val_batch)
    val_outs.append(out)
validation_epoch_end(val_outs)

Parameters

batch (Tensor | (Tensor, …) | [Tensor, …]) – The output of your DataLoader. A tensor, tuple or list.
batch_idx (int) – The index of this batch
dataloader_idx (int) – The index of the dataloader that produced this batch (only if multiple val dataloaders used)

Returns

Any of.

Any object or value

None - Validation will skip to the next batch

# pseudocode of order
val_outs = []
for val_batch in val_data:
    out = validation_step(val_batch)
    if defined('validation_step_end'):
        out = validation_step_end(out)
    val_outs.append(out)
val_outs = validation_epoch_end(val_outs)

# if you have one val dataloader:
def validation_step(self, batch, batch_idx)

# if you have multiple val dataloaders:
def validation_step(self, batch, batch_idx, dataloader_idx)

Examples:

# CASE 1: A single validation dataset
def validation_step(self, batch, batch_idx):
    x, y = batch

    # implement your own
    out = self(x)
    loss = self.loss(out, y)

    # log 6 example images
    # or generated text... or whatever
    sample_imgs = x[:6]
    grid = torchvision.utils.make_grid(sample_imgs)
    self.logger.experiment.add_image('example_images', grid, 0)

    # calculate acc
    labels_hat = torch.argmax(out, dim=1)
    val_acc = torch.sum(y == labels_hat).item() / (len(y) * 1.0)

    # log the outputs!
    self.log_dict({'val_loss': loss, 'val_acc': val_acc})

If you pass in multiple val dataloaders, validation_step() will have an additional argument.

# CASE 2: multiple validation dataloaders
def validation_step(self, batch, batch_idx, dataloader_idx):
    # dataloader_idx tells you which dataset this is.

Note

If you don’t need to validate you don’t need to implement this method.

Note

When the validation_step() is called, the model has been put in eval mode and PyTorch gradients have been disabled. At the end of validation, the model goes back to training mode and gradients are enabled.

cait.models.mh_predict(h5_path: str, feature_channel: int, group_name: str = 'events', prediction_name: str = 'prediction', model_handler: object = None, mh_path: str = None, which_data: str = 'mainpar')[source]¶

Add predictions from a Scikit-Learn model to the HDF5 set.

Parameters

h5_path (string) – The path to the HDF5 file.
feature_channel (int) – The channel of the detector module on that we make the predictions.
group_name (string) – The name of the group within the HDF5 file.
prediction_name (string) – The name of the prediction that is saved to the HDF5 set.
model_handler (object) – A model handler with that we want to make predictions.
mh_path (string) – A path to load a model handler from.
which_data (string) – Used for the evaluation tools instance, to tell which data is used for the prediction.

cait.models.nn_predict(h5_path: str, feature_channel: int, model: object = None, ptl_module: object = None, ptl_ckp_path: str = None, group_name: str = 'events', prediction_name: str = 'prediction', keys: list = ['event'], chunk_size: int = 50, no_channel_idx_in_pred: bool = False)[source]¶

Add predictions from a PyTorch model to the HDF5 set.

Parameters

h5_path (string) – The path to the HDF5 file.
feature_channel (int) – The channel of the detector module on that we make the predictions.
model (object) – A trained PyTorch Lightning module or Pytorch model.
ptl_module (object) – A Pytorch lightning model class.
ptl_ckp_path (string) – A path to the checkpoint from where we load the PyTorch lightning model parameters.
group_name (string) – The name of the group within the HDF5 file.
prediction_name (string) – The name of the prediction that is saved to the HDF5 set.
keys (list) – The keys from the HDF5 set that are included as features into every sample handed to the neural network.
chunk_size (int) – The size of the chunks to predict at once.
no_channel_idx_in_pred (bool) – If True, then we assume that there is no channel index in the data set from the HDF5 file.