CODEWITHSUNDEEP
pythonmachine-learningkerasdata-science
pr338
pr338

Reputation: 9120

How to log data about model expected performance in Keras?

I built a script that saves my best model to the filesystem and I want it to also log data about the expected performance of the model to a different text file.

Current code:

model = Sequential()

model.add(Dense(64, input_dim=14, init='uniform'))
model.add(LeakyReLU(alpha=0.3))
model.add(BatchNormalization(epsilon=1e-06, mode=0, momentum=0.9, weights=None))
model.add(Dropout(0.5))

model.add(Dense(64, init='uniform'))
model.add(LeakyReLU(alpha=0.3))
model.add(BatchNormalization(epsilon=1e-06, mode=0, momentum=0.9, weights=None))
model.add(Dropout(0.5))

model.add(Dense(2, init='uniform'))
model.add(Activation('softmax'))


sgd = SGD(lr=0.1, decay=1e-6, momentum=0.9, nesterov=True)
model.compile(loss='binary_crossentropy', optimizer=sgd)

checkpointer = ModelCheckpoint(filepath="/Desktop/SquareSpace/checkpointer", verbose=2, save_best_only=True)
model.fit(X_train, y_train, nb_epoch=100, batch_size=50, show_accuracy=True, validation_split=0.2, verbose = 2, callbacks=[checkpointer])

I currently save my best model as an HDF5 file. Is there a way to log the performance of the best model which is saved in the HDF5 file using Keras in a different text file?

EDIT: Is there a way to use this here http://keras.io/callbacks/?

Upvotes: 1

Views: 567

Answers (1)

Haifeng Zhang
Haifeng Zhang

Reputation: 31895

You can use python logging module to log your data.python logging

Use FileHandler to write all your data into a file and also you can use ConsoleHandler to display all data on console.

TimedRotatingFileHandler can help you rotate your log file in certain time intervals. For instance: every hour, every day or every week.

Upvotes: 2

Related Questions