Ta strona została przetłumaczona przez Cloud Translation API.
Switch to English

Pisanie własnych callbacków

Zobacz na TensorFlow.org Uruchom w Google Colab Wyświetl źródło na GitHub Pobierz notatnik

Wprowadzenie

Wywołanie zwrotne to potężne narzędzie do dostosowywania zachowania modelu Keras podczas uczenia, oceny lub wnioskowania. Przykłady obejmują tf.keras.callbacks.TensorBoard do wizualizacji postępu i wyników treningu za pomocą TensorBoard lub tf.keras.callbacks.ModelCheckpoint do okresowego zapisywania modelu podczas treningu.

W tym przewodniku dowiesz się, czym jest wywołanie zwrotne Keras, co może zrobić i jak zbudować własne. Udostępniamy kilka demonstracji prostych aplikacji zwrotnych, abyś mógł zacząć.

Ustawiać

 import tensorflow as tf
from tensorflow import keras
 

Omówienie wywołań zwrotnych Keras

Wszystkie wywołania zwrotne stanowią podklasę klasy keras.callbacks.Callback i zastępują zestaw metod wywoływanych na różnych etapach uczenia, testowania i przewidywania. Wywołania zwrotne są przydatne, aby uzyskać wgląd w wewnętrzne stany i statystyki modelu podczas uczenia.

Możesz przekazać listę wywołań zwrotnych (jako argument callbacks ) do następujących metod modelu:

Omówienie metod wywołania zwrotnego

Metody globalne

on_(train|test|predict)_begin(self, logs=None)

Nazywany na początku fit / evaluate / predict .

on_(train|test|predict)_end(self, logs=None)

Nazywany pod koniec fit / evaluate / predict .

Metody szkolenia / testowania / przewidywania na poziomie partii

on_(train|test|predict)_batch_begin(self, batch, logs=None)

Wywoływane tuż przed przetworzeniem partii podczas uczenia / testowania / przewidywania.

on_(train|test|predict)_batch_end(self, batch, logs=None)

Wywoływane pod koniec szkolenia / testowania / przewidywania partii. W ramach tej metody logs są dyktami zawierającymi wyniki metryk.

Metody na poziomie epoki (tylko szkolenie)

on_epoch_begin(self, epoch, logs=None)

Nazywany na początku epoki podczas treningu.

on_epoch_end(self, epoch, logs=None)

Nazywany pod koniec epoki podczas treningu.

Podstawowy przykład

Spójrzmy na konkretny przykład. Aby rozpocząć, zaimportujmy tensorflow i zdefiniujmy prosty model Sequential Keras:

 # Define the Keras model to add callbacks to
def get_model():
    model = keras.Sequential()
    model.add(keras.layers.Dense(1, input_dim=784))
    model.compile(
        optimizer=keras.optimizers.RMSprop(learning_rate=0.1),
        loss="mean_squared_error",
        metrics=["mean_absolute_error"],
    )
    return model

 

Następnie załaduj dane MNIST do uczenia i testowania z interfejsu API zestawów danych Keras:

 # Load example MNIST data and pre-process it
(x_train, y_train), (x_test, y_test) = tf.keras.datasets.mnist.load_data()
x_train = x_train.reshape(-1, 784).astype("float32") / 255.0
x_test = x_test.reshape(-1, 784).astype("float32") / 255.0

# Limit the data to 1000 samples
x_train = x_train[:1000]
y_train = y_train[:1000]
x_test = x_test[:1000]
y_test = y_test[:1000]
 
Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/mnist.npz
11493376/11490434 [==============================] - 0s 0us/step

Teraz zdefiniuj proste niestandardowe wywołanie zwrotne, które rejestruje:

  • Kiedy fit / evaluate / predict początek i koniec
  • Kiedy każda epoka zaczyna się i kończy
  • Kiedy zaczyna się i kończy każda partia treningowa
  • Kiedy rozpoczyna się i kończy każda partia oceny (test)
  • Kiedy każda partia wnioskowania (przewidywania) zaczyna się i kończy
 class CustomCallback(keras.callbacks.Callback):
    def on_train_begin(self, logs=None):
        keys = list(logs.keys())
        print("Starting training; got log keys: {}".format(keys))

    def on_train_end(self, logs=None):
        keys = list(logs.keys())
        print("Stop training; got log keys: {}".format(keys))

    def on_epoch_begin(self, epoch, logs=None):
        keys = list(logs.keys())
        print("Start epoch {} of training; got log keys: {}".format(epoch, keys))

    def on_epoch_end(self, epoch, logs=None):
        keys = list(logs.keys())
        print("End epoch {} of training; got log keys: {}".format(epoch, keys))

    def on_test_begin(self, logs=None):
        keys = list(logs.keys())
        print("Start testing; got log keys: {}".format(keys))

    def on_test_end(self, logs=None):
        keys = list(logs.keys())
        print("Stop testing; got log keys: {}".format(keys))

    def on_predict_begin(self, logs=None):
        keys = list(logs.keys())
        print("Start predicting; got log keys: {}".format(keys))

    def on_predict_end(self, logs=None):
        keys = list(logs.keys())
        print("Stop predicting; got log keys: {}".format(keys))

    def on_train_batch_begin(self, batch, logs=None):
        keys = list(logs.keys())
        print("...Training: start of batch {}; got log keys: {}".format(batch, keys))

    def on_train_batch_end(self, batch, logs=None):
        keys = list(logs.keys())
        print("...Training: end of batch {}; got log keys: {}".format(batch, keys))

    def on_test_batch_begin(self, batch, logs=None):
        keys = list(logs.keys())
        print("...Evaluating: start of batch {}; got log keys: {}".format(batch, keys))

    def on_test_batch_end(self, batch, logs=None):
        keys = list(logs.keys())
        print("...Evaluating: end of batch {}; got log keys: {}".format(batch, keys))

    def on_predict_batch_begin(self, batch, logs=None):
        keys = list(logs.keys())
        print("...Predicting: start of batch {}; got log keys: {}".format(batch, keys))

    def on_predict_batch_end(self, batch, logs=None):
        keys = list(logs.keys())
        print("...Predicting: end of batch {}; got log keys: {}".format(batch, keys))

 

Wypróbujmy to:

 model = get_model()
model.fit(
    x_train,
    y_train,
    batch_size=128,
    epochs=1,
    verbose=0,
    validation_split=0.5,
    callbacks=[CustomCallback()],
)

res = model.evaluate(
    x_test, y_test, batch_size=128, verbose=0, callbacks=[CustomCallback()]
)

res = model.predict(x_test, batch_size=128, callbacks=[CustomCallback()])
 
Starting training; got log keys: []
Start epoch 0 of training; got log keys: []
...Training: start of batch 0; got log keys: []
...Training: end of batch 0; got log keys: ['loss', 'mean_absolute_error']
...Training: start of batch 1; got log keys: []
...Training: end of batch 1; got log keys: ['loss', 'mean_absolute_error']
...Training: start of batch 2; got log keys: []
...Training: end of batch 2; got log keys: ['loss', 'mean_absolute_error']
...Training: start of batch 3; got log keys: []
...Training: end of batch 3; got log keys: ['loss', 'mean_absolute_error']
Start testing; got log keys: []
...Evaluating: start of batch 0; got log keys: []
...Evaluating: end of batch 0; got log keys: ['loss', 'mean_absolute_error']
...Evaluating: start of batch 1; got log keys: []
...Evaluating: end of batch 1; got log keys: ['loss', 'mean_absolute_error']
...Evaluating: start of batch 2; got log keys: []
...Evaluating: end of batch 2; got log keys: ['loss', 'mean_absolute_error']
...Evaluating: start of batch 3; got log keys: []
...Evaluating: end of batch 3; got log keys: ['loss', 'mean_absolute_error']
Stop testing; got log keys: []
End epoch 0 of training; got log keys: ['loss', 'mean_absolute_error', 'val_loss', 'val_mean_absolute_error']
Stop training; got log keys: []
Start testing; got log keys: []
...Evaluating: start of batch 0; got log keys: []
...Evaluating: end of batch 0; got log keys: ['loss', 'mean_absolute_error']
...Evaluating: start of batch 1; got log keys: []
...Evaluating: end of batch 1; got log keys: ['loss', 'mean_absolute_error']
...Evaluating: start of batch 2; got log keys: []
...Evaluating: end of batch 2; got log keys: ['loss', 'mean_absolute_error']
...Evaluating: start of batch 3; got log keys: []
...Evaluating: end of batch 3; got log keys: ['loss', 'mean_absolute_error']
...Evaluating: start of batch 4; got log keys: []
...Evaluating: end of batch 4; got log keys: ['loss', 'mean_absolute_error']
...Evaluating: start of batch 5; got log keys: []
...Evaluating: end of batch 5; got log keys: ['loss', 'mean_absolute_error']
...Evaluating: start of batch 6; got log keys: []
...Evaluating: end of batch 6; got log keys: ['loss', 'mean_absolute_error']
...Evaluating: start of batch 7; got log keys: []
...Evaluating: end of batch 7; got log keys: ['loss', 'mean_absolute_error']
Stop testing; got log keys: []
Start predicting; got log keys: []
...Predicting: start of batch 0; got log keys: []
...Predicting: end of batch 0; got log keys: ['outputs']
...Predicting: start of batch 1; got log keys: []
...Predicting: end of batch 1; got log keys: ['outputs']
...Predicting: start of batch 2; got log keys: []
...Predicting: end of batch 2; got log keys: ['outputs']
...Predicting: start of batch 3; got log keys: []
...Predicting: end of batch 3; got log keys: ['outputs']
...Predicting: start of batch 4; got log keys: []
...Predicting: end of batch 4; got log keys: ['outputs']
...Predicting: start of batch 5; got log keys: []
...Predicting: end of batch 5; got log keys: ['outputs']
...Predicting: start of batch 6; got log keys: []
...Predicting: end of batch 6; got log keys: ['outputs']
...Predicting: start of batch 7; got log keys: []
...Predicting: end of batch 7; got log keys: ['outputs']
Stop predicting; got log keys: []

Korzystanie z logs dyktuje

Dykt logs zawiera wartość straty i wszystkie metryki na końcu partii lub epoki. Przykład obejmuje stratę i średni błąd bezwzględny.

 class LossAndErrorPrintingCallback(keras.callbacks.Callback):
    def on_train_batch_end(self, batch, logs=None):
        print("For batch {}, loss is {:7.2f}.".format(batch, logs["loss"]))

    def on_test_batch_end(self, batch, logs=None):
        print("For batch {}, loss is {:7.2f}.".format(batch, logs["loss"]))

    def on_epoch_end(self, epoch, logs=None):
        print(
            "The average loss for epoch {} is {:7.2f} "
            "and mean absolute error is {:7.2f}.".format(
                epoch, logs["loss"], logs["mean_absolute_error"]
            )
        )


model = get_model()
model.fit(
    x_train,
    y_train,
    batch_size=128,
    epochs=2,
    verbose=0,
    callbacks=[LossAndErrorPrintingCallback()],
)

res = model.evaluate(
    x_test,
    y_test,
    batch_size=128,
    verbose=0,
    callbacks=[LossAndErrorPrintingCallback()],
)
 
For batch 0, loss is   30.07.
For batch 1, loss is  413.20.
For batch 2, loss is  284.62.
For batch 3, loss is  215.74.
For batch 4, loss is  174.43.
For batch 5, loss is  146.54.
For batch 6, loss is  126.53.
For batch 7, loss is  114.00.
The average loss for epoch 0 is  114.00 and mean absolute error is    5.88.
For batch 0, loss is    5.26.
For batch 1, loss is    5.16.
For batch 2, loss is    4.90.
For batch 3, loss is    4.91.
For batch 4, loss is    4.82.
For batch 5, loss is    4.64.
For batch 6, loss is    4.51.
For batch 7, loss is    4.52.
The average loss for epoch 1 is    4.52 and mean absolute error is    1.71.
For batch 0, loss is    7.47.
For batch 1, loss is    7.51.
For batch 2, loss is    7.33.
For batch 3, loss is    7.34.
For batch 4, loss is    7.35.
For batch 5, loss is    7.48.
For batch 6, loss is    7.44.
For batch 7, loss is    7.36.

Wykorzystanie atrybutu self.model

Oprócz otrzymywania informacji dziennika, gdy wywoływana jest jedna z ich metod, wywołania zwrotne mają dostęp do modelu skojarzonego z bieżącą rundą uczenia / oceny / wnioskowania: self.model .

Oto kilka rzeczy, które możesz zrobić z self.model w wywołaniu zwrotnym:

  • Ustaw self.model.stop_training = True aby natychmiast przerwać trening.
  • Zmutuj hiperparametry optymalizatora (dostępne jako self.model.optimizer ), takie jak self.model.optimizer.learning_rate .
  • Zapisz model w odstępach czasu.
  • Zapisz wynik działania model.predict() na kilku próbkach testowych pod koniec każdej epoki, aby użyć go jako testu poczytalności podczas uczenia.
  • Wyodrębnij wizualizacje cech pośrednich pod koniec każdej epoki, aby monitorować, czego model uczy się w czasie.
  • itp.

Zobaczmy to w akcji na kilku przykładach.

Przykłady aplikacji wywołania zwrotnego Keras

Wczesne zatrzymanie przy minimalnej stracie

Ten pierwszy przykład pokazuje tworzenie Callback które zatrzymuje uczenie po osiągnięciu minimum strat, ustawiając atrybut self.model.stop_training (boolean). Opcjonalnie możesz podać patience argumentacji, aby określić, ile epok powinniśmy odczekać przed zatrzymaniem po osiągnięciu lokalnego minimum.

tf.keras.callbacks.EarlyStopping zapewnia bardziej kompletną i ogólną implementację.

 import numpy as np


class EarlyStoppingAtMinLoss(keras.callbacks.Callback):
    """Stop training when the loss is at its min, i.e. the loss stops decreasing.

  Arguments:
      patience: Number of epochs to wait after min has been hit. After this
      number of no improvement, training stops.
  """

    def __init__(self, patience=0):
        super(EarlyStoppingAtMinLoss, self).__init__()
        self.patience = patience
        # best_weights to store the weights at which the minimum loss occurs.
        self.best_weights = None

    def on_train_begin(self, logs=None):
        # The number of epoch it has waited when loss is no longer minimum.
        self.wait = 0
        # The epoch the training stops at.
        self.stopped_epoch = 0
        # Initialize the best as infinity.
        self.best = np.Inf

    def on_epoch_end(self, epoch, logs=None):
        current = logs.get("loss")
        if np.less(current, self.best):
            self.best = current
            self.wait = 0
            # Record the best weights if current results is better (less).
            self.best_weights = self.model.get_weights()
        else:
            self.wait += 1
            if self.wait >= self.patience:
                self.stopped_epoch = epoch
                self.model.stop_training = True
                print("Restoring model weights from the end of the best epoch.")
                self.model.set_weights(self.best_weights)

    def on_train_end(self, logs=None):
        if self.stopped_epoch > 0:
            print("Epoch %05d: early stopping" % (self.stopped_epoch + 1))


model = get_model()
model.fit(
    x_train,
    y_train,
    batch_size=64,
    steps_per_epoch=5,
    epochs=30,
    verbose=0,
    callbacks=[LossAndErrorPrintingCallback(), EarlyStoppingAtMinLoss()],
)
 
For batch 0, loss is   27.80.
For batch 1, loss is  392.24.
For batch 2, loss is  269.56.
For batch 3, loss is  205.69.
For batch 4, loss is  166.82.
The average loss for epoch 0 is  166.82 and mean absolute error is    7.96.
For batch 0, loss is    7.27.
For batch 1, loss is    6.18.
For batch 2, loss is    6.08.
For batch 3, loss is    5.83.
For batch 4, loss is    5.61.
The average loss for epoch 1 is    5.61 and mean absolute error is    1.91.
For batch 0, loss is    5.03.
For batch 1, loss is    6.90.
For batch 2, loss is    7.60.
For batch 3, loss is    7.69.
For batch 4, loss is    8.49.
The average loss for epoch 2 is    8.49 and mean absolute error is    2.41.
Restoring model weights from the end of the best epoch.
Epoch 00003: early stopping

<tensorflow.python.keras.callbacks.History at 0x7fa32a9822e8>

Planowanie kursu nauki

W tym przykładzie pokazujemy, jak niestandardowy Callback może być użyty do dynamicznej zmiany szybkości uczenia się optymalizatora w trakcie szkolenia.

Zobacz callbacks.LearningRateScheduler aby callbacks.LearningRateScheduler z bardziej ogólnymi implementacjami.

 class CustomLearningRateScheduler(keras.callbacks.Callback):
    """Learning rate scheduler which sets the learning rate according to schedule.

  Arguments:
      schedule: a function that takes an epoch index
          (integer, indexed from 0) and current learning rate
          as inputs and returns a new learning rate as output (float).
  """

    def __init__(self, schedule):
        super(CustomLearningRateScheduler, self).__init__()
        self.schedule = schedule

    def on_epoch_begin(self, epoch, logs=None):
        if not hasattr(self.model.optimizer, "lr"):
            raise ValueError('Optimizer must have a "lr" attribute.')
        # Get the current learning rate from model's optimizer.
        lr = float(tf.keras.backend.get_value(self.model.optimizer.learning_rate))
        # Call schedule function to get the scheduled learning rate.
        scheduled_lr = self.schedule(epoch, lr)
        # Set the value back to the optimizer before this epoch starts
        tf.keras.backend.set_value(self.model.optimizer.lr, scheduled_lr)
        print("\nEpoch %05d: Learning rate is %6.4f." % (epoch, scheduled_lr))


LR_SCHEDULE = [
    # (epoch to start, learning rate) tuples
    (3, 0.05),
    (6, 0.01),
    (9, 0.005),
    (12, 0.001),
]


def lr_schedule(epoch, lr):
    """Helper function to retrieve the scheduled learning rate based on epoch."""
    if epoch < LR_SCHEDULE[0][0] or epoch > LR_SCHEDULE[-1][0]:
        return lr
    for i in range(len(LR_SCHEDULE)):
        if epoch == LR_SCHEDULE[i][0]:
            return LR_SCHEDULE[i][1]
    return lr


model = get_model()
model.fit(
    x_train,
    y_train,
    batch_size=64,
    steps_per_epoch=5,
    epochs=15,
    verbose=0,
    callbacks=[
        LossAndErrorPrintingCallback(),
        CustomLearningRateScheduler(lr_schedule),
    ],
)
 

Epoch 00000: Learning rate is 0.1000.
For batch 0, loss is   36.98.
For batch 1, loss is  505.58.
For batch 2, loss is  344.37.
For batch 3, loss is  260.42.
For batch 4, loss is  210.08.
The average loss for epoch 0 is  210.08 and mean absolute error is    8.70.

Epoch 00001: Learning rate is 0.1000.
For batch 0, loss is    6.16.
For batch 1, loss is    5.91.
For batch 2, loss is    5.25.
For batch 3, loss is    5.30.
For batch 4, loss is    5.15.
The average loss for epoch 1 is    5.15 and mean absolute error is    1.84.

Epoch 00002: Learning rate is 0.1000.
For batch 0, loss is    8.28.
For batch 1, loss is    7.63.
For batch 2, loss is    6.94.
For batch 3, loss is    6.69.
For batch 4, loss is    6.24.
The average loss for epoch 2 is    6.24 and mean absolute error is    1.98.

Epoch 00003: Learning rate is 0.0500.
For batch 0, loss is    6.48.
For batch 1, loss is    5.40.
For batch 2, loss is    4.70.
For batch 3, loss is    4.45.
For batch 4, loss is    4.08.
The average loss for epoch 3 is    4.08 and mean absolute error is    1.64.

Epoch 00004: Learning rate is 0.0500.
For batch 0, loss is    3.30.
For batch 1, loss is    4.07.
For batch 2, loss is    4.14.
For batch 3, loss is    4.02.
For batch 4, loss is    4.07.
The average loss for epoch 4 is    4.07 and mean absolute error is    1.62.

Epoch 00005: Learning rate is 0.0500.
For batch 0, loss is    4.81.
For batch 1, loss is    5.43.
For batch 2, loss is    4.83.
For batch 3, loss is    4.90.
For batch 4, loss is    4.58.
The average loss for epoch 5 is    4.58 and mean absolute error is    1.73.

Epoch 00006: Learning rate is 0.0100.
For batch 0, loss is    2.87.
For batch 1, loss is    3.13.
For batch 2, loss is    2.93.
For batch 3, loss is    3.02.
For batch 4, loss is    3.08.
The average loss for epoch 6 is    3.08 and mean absolute error is    1.41.

Epoch 00007: Learning rate is 0.0100.
For batch 0, loss is    3.33.
For batch 1, loss is    4.50.
For batch 2, loss is    4.32.
For batch 3, loss is    4.17.
For batch 4, loss is    3.97.
The average loss for epoch 7 is    3.97 and mean absolute error is    1.57.

Epoch 00008: Learning rate is 0.0100.
For batch 0, loss is    3.20.
For batch 1, loss is    3.54.
For batch 2, loss is    3.16.
For batch 3, loss is    3.29.
For batch 4, loss is    3.54.
The average loss for epoch 8 is    3.54 and mean absolute error is    1.47.

Epoch 00009: Learning rate is 0.0050.
For batch 0, loss is    3.39.
For batch 1, loss is    3.00.
For batch 2, loss is    3.12.
For batch 3, loss is    3.36.
For batch 4, loss is    3.20.
The average loss for epoch 9 is    3.20 and mean absolute error is    1.43.

Epoch 00010: Learning rate is 0.0050.
For batch 0, loss is    3.72.
For batch 1, loss is    3.55.
For batch 2, loss is    3.21.
For batch 3, loss is    2.98.
For batch 4, loss is    3.02.
The average loss for epoch 10 is    3.02 and mean absolute error is    1.36.

Epoch 00011: Learning rate is 0.0050.
For batch 0, loss is    2.88.
For batch 1, loss is    2.88.
For batch 2, loss is    2.89.
For batch 3, loss is    2.95.
For batch 4, loss is    3.50.
The average loss for epoch 11 is    3.50 and mean absolute error is    1.44.

Epoch 00012: Learning rate is 0.0010.
For batch 0, loss is    3.26.
For batch 1, loss is    3.15.
For batch 2, loss is    3.59.
For batch 3, loss is    3.46.
For batch 4, loss is    3.44.
The average loss for epoch 12 is    3.44 and mean absolute error is    1.44.

Epoch 00013: Learning rate is 0.0010.
For batch 0, loss is    3.09.
For batch 1, loss is    3.34.
For batch 2, loss is    3.34.
For batch 3, loss is    3.31.
For batch 4, loss is    3.13.
The average loss for epoch 13 is    3.13 and mean absolute error is    1.43.

Epoch 00014: Learning rate is 0.0010.
For batch 0, loss is    3.69.
For batch 1, loss is    3.23.
For batch 2, loss is    3.09.
For batch 3, loss is    3.39.
For batch 4, loss is    3.19.
The average loss for epoch 14 is    3.19 and mean absolute error is    1.37.

<tensorflow.python.keras.callbacks.History at 0x7fa31c4a5c18>

Wbudowane wywołania zwrotne Keras

Koniecznie sprawdź istniejące wywołania zwrotne Keras, czytając dokumentację API . Aplikacje obejmują logowanie do pliku CSV, zapisywanie modelu, wizualizację metryk w TensorBoard i wiele więcej!