"Add parameter 'dataset_sink_mode' for RDPMonitor class."

mindarmour/diff_privacy/mechanisms/mechanisms.py

@@ -86,7 +86,7 @@ class GaussianRandom(Mechanisms):
         >>> shape = (3, 2, 4)
         >>> norm_bound = 1.0
         >>> initial_noise_multiplier = 1.5
-        >>> net = GaussianRandom(shape, norm_bound, initial_noise_multiplier)
+        >>> net = GaussianRandom(norm_bound, initial_noise_multiplier)
         >>> res = net(shape)
         >>> print(res)
     """

mindarmour/diff_privacy/monitor/monitor.py

@@ -90,6 +90,8 @@ class RDPMonitor(Callback):
             training. Default: 6e-4.
         per_print_times　(int): The interval steps of computing and printing
             the privacy budget. Default: 50.
+        dataset_sink_mode (bool): If True, all training data would be passed to device(Ascend) at once. If False,
+            training data would be passed to device after each step training. Default: False.
 
     Examples:
         >>> rdp = PrivacyMonitorFactory.create(policy='rdp',
@@ -104,7 +106,7 @@ class RDPMonitor(Callback):
     def __init__(self, num_samples, batch_size, initial_noise_multiplier=1.5,
                  max_eps=10.0, target_delta=1e-3, max_delta=None,
                  target_eps=None, orders=None, noise_decay_mode='Time',
-                 noise_decay_rate=6e-4, per_print_times=50):
+                 noise_decay_rate=6e-4, per_print_times=50, dataset_sink_mode=False):
         super(RDPMonitor, self).__init__()
         check_int_positive('num_samples', num_samples)
         check_int_positive('batch_size', batch_size)
@@ -141,6 +143,7 @@ class RDPMonitor(Callback):
             noise_decay_rate = check_param_type('noise_decay_rate', noise_decay_rate, float)
             check_param_in_range('noise_decay_rate', noise_decay_rate, 0.0, 1.0)
         check_int_positive('per_print_times', per_print_times)
+        check_param_type('dataset_sink_mode', dataset_sink_mode, bool)
 
         self._total_echo_privacy = None
         self._num_samples = num_samples
@@ -155,6 +158,8 @@ class RDPMonitor(Callback):
         self._noise_decay_rate = noise_decay_rate
         self._rdp = 0
         self._per_print_times = per_print_times
+        if dataset_sink_mode:
+            self._per_print_times = int(self._num_samples / self._batch_size)
 
     def max_epoch_suggest(self):
         """