TensorFlowアドオンのコールバック：TimeStopping

TensorFlow.orgで表示 Google Colab で実行 GitHub でソースを表示{ ノートブックをダウンロード/a0}

概要

このノートブックでは、TensorFlowアドオンでTimeStoppingコールバックを使用する方法を紹介します。

セットアップ

pip install -q -U tensorflow-addons

import tensorflow_addons as tfa

from tensorflow.keras.datasets import mnist
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Dropout, Flatten

データのインポートと正規化

# the data, split between train and test sets
(x_train, y_train), (x_test, y_test) = mnist.load_data()
# normalize data
x_train, x_test = x_train / 255.0, x_test / 255.0

シンプルなMNIST CNNモデルの構築

# build the model using the Sequential API
model = Sequential()
model.add(Flatten(input_shape=(28, 28)))
model.add(Dense(128, activation='relu'))
model.add(Dropout(0.2))
model.add(Dense(10, activation='softmax'))

model.compile(optimizer='adam',
              loss = 'sparse_categorical_crossentropy',
              metrics=['accuracy'])

シンプルなTimeStoppingの使用法

# initialize TimeStopping callback 
time_stopping_callback = tfa.callbacks.TimeStopping(seconds=5, verbose=1)

# train the model with tqdm_callback
# make sure to set verbose = 0 to disable
# the default progress bar.
model.fit(x_train, y_train,
          batch_size=64,
          epochs=100,
          callbacks=[time_stopping_callback],
          validation_data=(x_test, y_test))

Epoch 1/100
938/938 [==============================] - 3s 3ms/step - loss: 0.5543 - accuracy: 0.8400 - val_loss: 0.1561 - val_accuracy: 0.9564
Epoch 2/100
938/938 [==============================] - 2s 2ms/step - loss: 0.1711 - accuracy: 0.9504 - val_loss: 0.1126 - val_accuracy: 0.9669
Timed stopping at epoch 2 after training for 0:00:05
<tensorflow.python.keras.callbacks.History at 0x7fb6af066240>