I could not understand that DNNClassifier prints progress during training, i.e. loss assessment and validation. Since I realized that the loss can be printed using a configuration parameter that inherits from BaseEstimator, but when I passed the RunConfig object, the classifier did not print anything.
from tensorflow.contrib.learn.python.learn.estimators import run_config config = run_config.RunConfig(verbose=1) classifier = learn.DNNClassifier(hidden_units=[10, 20, 10], n_classes=3, config=config) classifier.fit(X_train, y_train, steps=1000)
Am I missing something? I checked how RunConfig handles the detailed parameter, and it seems that it only cares that there are more than 1 , which does not correspond to the documentation:
verbose: Controls detail, possible values: 0: algorithm and debugging information are disabled. 1: coach prints progress. 2: the device log is printed.
Regarding the validation assessment, I thought using monitors. ValidationMonitor will be just fine, but when you try it, the classifier does not print anything, and nothing happens when you try to use early_stopping_rounds. I am looking for documentation or some comments in the source code, but I could not find them for monitors.
source share