Is the following way the right usage to enable profiler with tf estimator?
tf.profiler.experimental.start(FLAGS.working_dir)
tf.estimator.train_and_evaluate(
classifier,
train_spec=tf.estimator.TrainSpec(input_fn=input_fn, max_steps=FLAGS.steps, hooks=hooks),
eval_spec=tf.estimator.EvalSpec(input_fn=input_fn)
)
tf.profiler.experimental.stop()
However the profiling result doesn't include any steps:
