paddlets.models.common.callbacks.callbacks

class Callback

Bases: object

Abstract base class used to build new callbacks.

_trainer

A model instance.

Type

PaddleBaseModel

set_trainer(model: PaddleBaseModel)

Set model instance.

Parameters

model (PaddleBaseModel) – A model instance.

on_epoch_begin(epoch: int, logs: Optional[Dict[str, Any]] = None)

Called at the beginning of each epoch.

Parameters

• epoch (int) – The index of epoch.

• logs (Dict[str, Any]|None) – The logs is a dict or None.

on_epoch_end(epoch: int, logs: Optional[Dict[str, Any]] = None)

Called at the end of each epoch.

Parameters

• epoch (int) – The index of epoch.

• logs (Dict[str, Any]|None) – The logs is a dict or None. contains loss and metrics.

on_batch_begin(batch: int, logs: Optional[Dict[str, Any]] = None)

Called at the beginning of each batch in training.

Parameters

• batch (int) – The index of batch.

• logs (Dict[str, Any]|None) – The logs is a dict or None.

on_batch_end(batch: int, logs: Optional[Dict[str, Any]] = None)

Called at the end of each batch in training.

Parameters

• batch (int) – The index of batch.

• logs (Dict[str, Any]|None) – The logs is a dict or None. contains loss and batch_size.

on_train_begin(logs: Optional[Dict[str, Any]] = None)

Called at the start of training.

Parameters

logs (Dict[str, Any]|None) – The logs is a dict or None.

on_train_end(logs: Optional[Dict[str, Any]] = None)

Called at the end of training.

Parameters

logs (Dict[str, Any]|None) – The logs is a dict or None.

class CallbackContainer(callbacks: List[Callback])

Bases: object

Container holding a list of callbacks.

Parameters

callbacks (List[Callback]) – List of callbacks.

_callbacks

List of callbacks.

Type

List[Callback]

append(callback: Callback)

Append callback to the container.

Parameters

callback (Callback) – Callback instance.

set_trainer(model: PaddleBaseModel)

Set model instance.

Parameters

model (PaddleBaseModel) – A model instance.

on_epoch_begin(epoch: int, logs: Optional[Dict[str, Any]] = None)

Called at the beginning of each epoch.

Parameters

• epoch (int) – The index of epoch.

• logs (Dict[str, Any]|None) – The logs is a dict or None.

on_epoch_end(epoch: int, logs: Optional[Dict[str, Any]] = None)

Called at the end of each epoch.

Parameters

• epoch (int) – The index of epoch.

• logs (Dict[str, Any]|None) – The logs is a dict or None. contains loss and metrics.

on_batch_begin(batch: int, logs: Optional[Dict[str, Any]] = None)

Called at the beginning of each batch in training.

Parameters

• batch (int) – The index of batch.

• logs (Dict[str, Any]|None) – The logs is a dict or None.

on_batch_end(batch: int, logs: Optional[Dict[str, Any]] = None)

Called at the end of each batch in training.

Parameters

• batch (int) – The index of batch.

• logs (Dict[str, Any]|None) – The logs is a dict or None. contains loss and batch_size.

on_train_begin(logs: Optional[Dict[str, Any]] = None)

Called at the start of training.

Parameters

logs (Dict[str, Any]|None) – The logs is a dict or None.

on_train_end(logs: Optional[Dict[str, Any]] = None)

Called at the end of training.

Parameters

logs (Dict[str, Any]|None) – The logs is a dict or None.

class EarlyStopping(early_stopping_metric: str, is_maximize: bool, tol: float = 0.0, patience: int = 1)

Bases: Callback

EarlyStopping callback, allow the trainer to exit the training loop if the given metric stopped improving during evaluation.

Parameters

• early_stopping_metric (str) – Early stopping metric name.

• is_maximize (bool) – Whether to maximize or not early_stopping_metric.

• tol (float) – Minimum change in monitored value to qualify as improvement. This number should be positive.

• patience (int) – Number of epochs to wait for improvement before terminating. the counter be reset after each improvement

_early_stopping_metric

Early stopping metric name.

Type

str

_is_maximize

Whether to maximize or not early_stopping_metric.

Type

bool

_tol

Minimum change in monitored value to qualify as improvement.

Type

float

_patience

Number of epochs to wait for improvement before terminating.

Type

int

_best_epoch

Best epoch.

Type

int

_stopped_epoch

Stopped epoch.

Type

int

_best_loss

Best loss.

Type

float

_wait

Number of times that the early_stopping_metric failed to improve.

Type

int

on_epoch_end(epoch: int, logs: Optional[Dict[str, Any]] = None)

Called at the end of each epoch.

Parameters

• epoch (int) – The index of epoch.

• logs (Dict[str, Any]|None) – The logs is a dict or None. contains loss and metrics.

on_train_end(logs: Optional[Dict[str, Any]] = None)

Called at the end of training.

Parameters

logs (Dict[str, Any]|None) – The logs is a dict or None.

class History(verbose: int = 1)

Bases: Callback

Callback that records events into a History object.

Parameters

verbose (int) – Print results every verbose iteration.

_verbose

Print results every verbose iteration.

Type

int

_history

Record all information of metrics of each epoch.

Type

Dict[str, Any]

_start_time

Start time of training.

Type

float

_epoch_loss

Average loss per epoch.

Type

float

_epoch_metrics

Record all information of metrics of each epoch.

Type

Dict[str, Any]

_samples_seen

Traversed samples.

Type

int

on_train_begin(logs: Optional[Dict[str, Any]] = None)

Called at the start of training.

Parameters

logs (Dict[str, Any]|None) – The logs is a dict or None.

on_epoch_begin(epoch: int, logs: Optional[Dict[str, Any]] = None)

Called at the beginning of each epoch.

Parameters

• epoch (int) – The index of epoch.

• logs (Dict[str, Any]|None) – The logs is a dict or None.

on_epoch_end(epoch: int, logs: Optional[Dict[str, Any]] = None)

Called at the end of each epoch.

Parameters

• epoch (int) – The index of epoch.

• logs (Dict[str, Any]|None) – The logs is a dict or None. contains loss and metrics.

on_batch_end(batch: int, logs: Optional[Dict[str, Any]] = None)

Called at the end of each batch in training.

Parameters

• batch (int) – The index of batch.

• logs (Dict[str, Any]|None) – The logs is a dict or None. contains loss and batch_size.