CSVLogger

[source]

CSVLogger class

keras.callbacks.CSVLogger(filename, separator=",", append=False)

Callback that streams epoch results to a CSV file.

Supports all values that can be represented as a string, including 1D iterables such as np.ndarray.

Arguments

  • filename: Filename of the CSV file, e.g. 'run/log.csv'.
  • separator: String used to separate elements in the CSV file.
  • append: Boolean. True: append if file exists (useful for continuing training). False: overwrite existing file.

Example

csv_logger = CSVLogger('training.log')
model.fit(X_train, Y_train, callbacks=[csv_logger])