enchanter.tasks¶

Classification Task¶

ClassificationRunner¶

class enchanter.tasks.ClassificationRunner(model: torch.nn.modules.module.Module, optimizer: torch.optim.optimizer.Optimizer, criterion: torch.nn.modules.loss._Loss, experiment, scheduler: Optional[List] = None, callbacks: Optional[List[enchanter.callbacks.base.Callback]] = None)[source]¶

Bases: enchanter.engine.runner.BaseRunner

Runner for classification tasks.

Examples

>>> from comet_ml import Experiment
>>> import torch
>>> model = torch.nn.Sequential(...)
>>> optimizer = torch.optim.Adam(model.parameters())
>>> criterion = torch.nn.CrossEntropyLoss()
>>> runner = ClassificationRunner(
>>>     model,
>>>     optimizer,
>>>     criterion,
>>>     Experiment()
>>> )
>>> runner.fit(...)
>>> # or
>>> runner.add_loader(...)
>>> runner.train_config(epochs=10)
>>> runner.run()

Methods

`general_step`(batch)	This method is executed by train_step, val_step, test_step.
`general_end`(outputs)	This method is executed by train_end, val_end, test_end.

static general_end(outputs: List) → Dict[source]¶

This method is executed by train_end, val_end, test_end.

Parameters: outputs –

Returns:

general_step(batch: Tuple) → Dict[source]¶

This method is executed by train_step, val_step, test_step.

Parameters: batch –

Returns:

predict(x: Union[torch.Tensor, numpy.ndarray]) → numpy.ndarray[source]¶

A method that makes predictions based on the given input.

Parameters: x (Union[torch.Tensor, np.ndarray]) –
Returns: predict

test_end(outputs: List) → Dict[source]¶

This method is executed at the end of each step of neural network testing.

Parameters: outputs –
Returns: You need to return a dictionary.

test_step(batch: Tuple) → Dict[source]¶

This method is executed at every 1 step when testing the neural net. See train_step() for help.

Parameters: batch – A tuple containing data & labels get from the PyTorch DataLoader.
Returns: You need to return a dictionary.

train_end(outputs: List) → Dict[source]¶

This method is executed at the end of each step of neural network training.

Parameters: outputs –

Returns:

train_step(batch: Tuple) → Dict[source]¶

When training the neural network,

>>> import torch.nn as nn
>>> train_loader: DataLoader = ...
>>> model: nn.Module = ...
>>> criterion = ...
>>> for x, y in train_loader:
>>>     out = model(x)
>>>     loss = criterion(out, y)

this method is responsible for the above areas.

Parameters: batch – A tuple containing data & labels get from the PyTorch DataLoader.
Returns: You need to return a dictionary with the key ‘loss’.

Examples

>>> def train_step(self, batch):
>>>     x, y = batch
>>>     out = self.model(x)
>>>     loss = nn.functional.cross_entropy(out, y)
>>>     return {"loss": loss}

val_end(outputs: List) → Dict[source]¶

This method is executed at the end of each step of neural network validating.

Parameters: outputs –
Returns: You need to return a dictionary.

val_step(batch: Tuple) → Dict[source]¶

This method is executed at every 1 step when validating the neural net. See train_step() for help.

Parameters: batch – A tuple containing data & labels get from the PyTorch DataLoader.
Returns: You need to return a dictionary.

Regression Task¶

RegressionnRunner¶

class enchanter.tasks.RegressionRunner(model: torch.nn.modules.module.Module, optimizer: torch.optim.optimizer.Optimizer, criterion: torch.nn.modules.loss._Loss, experiment, scheduler: Optional[List] = None, callbacks: Optional[List[enchanter.callbacks.base.Callback]] = None)[source]¶

Bases: enchanter.engine.runner.BaseRunner

Runner for regression problems.

Examples

>>> runner = RegressionRunner(...)
>>> runner.add_loader("train", ...)
>>> runner.train_config(epochs=1)
>>> runner.run()
>>> # OR
>>> runner = RegressionRunner(...)
>>> runner.fit(x, y, epochs=1, batch_size=32)

Methods

`general_step`(batch)	This method is executed by train_step, val_step, test_step.
`general_end`(outputs)	This method is executed by train_end, val_end, test_end.

static general_end(outputs: List) → Dict[source]¶

This method is executed by train_end, val_end, test_end.

Parameters: outputs –

Returns:

general_step(batch: Tuple) → Dict[source]¶

This method is executed by train_step, val_step, test_step.

Parameters: batch –

Returns:

predict(x: Union[torch.Tensor, numpy.ndarray]) → numpy.ndarray[source]¶

A method that makes predictions based on the given input.

Parameters: x (Union[torch.Tensor, np.ndarray]) –
Returns: predict

test_end(outputs: List) → Dict[source]¶

This method is executed at the end of each step of neural network testing.

Parameters: outputs –
Returns: You need to return a dictionary.

test_step(batch: Tuple) → Dict[source]¶

This method is executed at every 1 step when testing the neural net. See train_step() for help.

Parameters: batch – A tuple containing data & labels get from the PyTorch DataLoader.
Returns: You need to return a dictionary.

train_end(outputs: List) → Dict[source]¶

This method is executed at the end of each step of neural network training.

Parameters: outputs –

Returns:

train_step(batch: Tuple) → Dict[source]¶

When training the neural network,

>>> import torch.nn as nn
>>> train_loader: DataLoader = ...
>>> model: nn.Module = ...
>>> criterion = ...
>>> for x, y in train_loader:
>>>     out = model(x)
>>>     loss = criterion(out, y)

this method is responsible for the above areas.

Parameters: batch – A tuple containing data & labels get from the PyTorch DataLoader.
Returns: You need to return a dictionary with the key ‘loss’.

Examples

>>> def train_step(self, batch):
>>>     x, y = batch
>>>     out = self.model(x)
>>>     loss = nn.functional.cross_entropy(out, y)
>>>     return {"loss": loss}

val_end(outputs: List) → Dict[source]¶

This method is executed at the end of each step of neural network validating.

Parameters: outputs –
Returns: You need to return a dictionary.

val_step(batch: Tuple) → Dict[source]¶

This method is executed at every 1 step when validating the neural net. See train_step() for help.

Parameters: batch – A tuple containing data & labels get from the PyTorch DataLoader.
Returns: You need to return a dictionary.

AutoEncoderRunner¶

class enchanter.tasks.AutoEncoderRunner(model: torch.nn.modules.module.Module, optimizer: torch.optim.optimizer.Optimizer, criterion: torch.nn.modules.loss._Loss, experiment, scheduler: Optional[List] = None, callbacks: Optional[List[enchanter.callbacks.base.Callback]] = None)[source]¶

Bases: enchanter.tasks.regression.RegressionRunner

Runner for training AutoEncoder.

Methods

`general_step`(batch)	This method is executed by train_step, val_step, test_step.
`general_end`(outputs)	This method is executed by train_end, val_end, test_end.

static general_end(outputs: List) → Dict[source]¶

This method is executed by train_end, val_end, test_end.

Parameters: outputs –

Returns:

general_step(batch: Tuple) → Dict[source]¶

This method is executed by train_step, val_step, test_step.

Parameters: batch –

Returns:

Time Series Representation Learning Task¶

TimeSeriesUnsupervisedRunner¶

class enchanter.tasks.TimeSeriesUnsupervisedRunner(model: torch.nn.modules.module.Module, optimizer: torch.optim.optimizer.Optimizer, experiment, n_negative_samples: int = 1, negative_penalty: int = 1, compared_len: Optional[int] = None, evaluator: Union[sklearn.linear_model._base.LinearRegression, sklearn.tree._classes.BaseDecisionTree, sklearn.svm._base.BaseLibSVM, sklearn.neighbors._base.NeighborsBase] = SVC(), save_memory: bool = False, scheduler: Optional[List] = None, callbacks: Optional[List[enchanter.callbacks.base.Callback]] = None)[source]¶

Bases: enchanter.engine.runner.BaseRunner

Runner for unsupervised time series representation learning.

Unsupervised representation learning for time series uses the the Unsupervised Triplet Loss proposed in NeurIPS 2019.

Paper: Unsupervised Scalable Representation Learning for Multivariate Time Series

Examples

>>> experiment = ...
>>> model: torch.nn.Module = ...
>>> optimizer: torch.optim.Optimizer = ...
>>> runner = TimeSeriesUnsupervisedRunner(model, optimizer, experiment)
>>> runner.add_loader("train", ...)
>>> runner.train_config(...)
>>> runner.run()

Initializer

Parameters

model – PyTorch model which outputting a fixed-length vector regardless of the length of the input series.
optimizer – PyTorch Optimizer
experiment – comet_ml.BaseExperiment or enchanter.callbacks.BaseLogger
n_negative_samples – Parameter K in the paper. The number of negative samples to be sampled during training.
negative_penalty – Coefficients that control how much negative values are valued.
compared_len – Maximum length of randomly chosen time series. (default None).
save_memory – If True, enables to save GPU memory.
scheduler – lr scheduler. use torch.optim.lr_scheduler
callbacks – List of callback.

Methods

`calculate_negative_loss_per_negative_sample`(…)	calculate negative loss per negative sample
`calculate_negative_loss`(positive_loss, …)	calculate negative loss using all negative samples.
`encode`(data)	Output encoded data.
`train_step`(batch)	When training the neural network,
`train_end`(outputs)	This method is executed at the end of each step of neural network training.

calculate_negative_loss(positive_loss, anchor_representation)[source]¶

calculate negative loss using all negative samples.

Parameters

positive_loss –
anchor_representation –

Returns

loss (torch.Tensor)

calculate_negative_loss_per_negative_sample(begin_neg: torch.Tensor, len_pos_neg: int, batch_size: int, negative_sample_step: int, anchor_representation: torch.Tensor, data: torch.Tensor) → torch.Tensor[source]¶

calculate negative loss per negative sample

Parameters

begin_neg –
len_pos_neg –
batch_size –
negative_sample_step –
anchor_representation –
data –

Returns

negative loss (torch.Tensor)

encode(data: Union[numpy.ndarray, torch.Tensor]) → Union[numpy.ndarray, torch.Tensor][source]¶

Output encoded data. The output data has the same data type as the input.

Parameters: data – data
Returns: encoded data

initialize() → None[source]¶

The method that prepares the Runner. If the variables required for execution, such as self.model, self.optimizer, self.experiment, etc., are not defined, the program will exit with an error message. If there are no problems, pass the model to the CPU or GPU.

Returns: None

predict(x: Union[torch.Tensor, numpy.ndarray]) → numpy.ndarray[source]¶: See Also self.encode

test_end(outputs: List) → Dict[str, torch.Tensor][source]¶

This method is executed at the end of each step of neural network testing.

Parameters: outputs –
Returns: You need to return a dictionary.

test_step(batch: Tuple) → Dict[str, torch.Tensor][source]¶

This method is executed at every 1 step when testing the neural net. See train_step() for help.

Parameters: batch – A tuple containing data & labels get from the PyTorch DataLoader.
Returns: You need to return a dictionary.

train_end(outputs: List) → Dict[str, torch.Tensor][source]¶

This method is executed at the end of each step of neural network training.

Parameters: outputs –

Returns:

train_step(batch) → Dict[str, torch.Tensor][source]¶

When training the neural network,

>>> import torch.nn as nn
>>> train_loader: DataLoader = ...
>>> model: nn.Module = ...
>>> criterion = ...
>>> for x, y in train_loader:
>>>     out = model(x)
>>>     loss = criterion(out, y)

this method is responsible for the above areas.

Parameters: batch – A tuple containing data & labels get from the PyTorch DataLoader.
Returns: You need to return a dictionary with the key ‘loss’.

Examples

>>> def train_step(self, batch):
>>>     x, y = batch
>>>     out = self.model(x)
>>>     loss = nn.functional.cross_entropy(out, y)
>>>     return {"loss": loss}

val_end(outputs: List) → Dict[str, torch.Tensor][source]¶

This method is executed at the end of each step of neural network validating.

Parameters: outputs –
Returns: You need to return a dictionary.

val_step(batch: Tuple) → Dict[str, torch.Tensor][source]¶

This method is executed at every 1 step when validating the neural net. See train_step() for help.

Parameters: batch – A tuple containing data & labels get from the PyTorch DataLoader.
Returns: You need to return a dictionary.

Ensemble¶

BaseEnsembleEstimator¶

class enchanter.tasks.BaseEnsembleEstimator(runners: List[enchanter.engine.runner.BaseRunner], mode: Optional[str] = None)[source]¶: Bases: sklearn.base.BaseEstimator

SoftEnsemble¶

class enchanter.tasks.SoftEnsemble(runners: List[enchanter.engine.runner.BaseRunner], mode: Optional[str] = None)[source]¶

Bases: enchanter.tasks.ensemble.BaseEnsembleEstimator

Ensemble that averages probabilities

HardEnsemble¶

class enchanter.tasks.HardEnsemble(runners: List[enchanter.engine.runner.BaseRunner])[source]¶

Bases: enchanter.tasks.ensemble.BaseEnsembleEstimator

Ensemble that takes a majority vote