caffe2.python.optimizer.AdadeltaOptimizer Class Reference

Inheritance diagram for caffe2.python.optimizer.AdadeltaOptimizer:

Public Member Functions
def	__init__ (self, alpha=0.01, epsilon=1e-4, decay=0.95, policy="fixed", sparse_dedup_aggregator=None, engine='', kwargs)
def	scale_learning_rate (self, scale)
Public Member Functions inherited from caffe2.python.optimizer.Optimizer
def	__init__ (self)
def	__call__ (self, net, param_init_net, param, grad=None)
def	get_cpu_blob_name (self, base_str, node_name='')
def	get_gpu_blob_name (self, base_str, gpu_id, node_name)
def	make_unique_blob_name (self, base_str)
def	build_lr (self, net, param_init_net, base_learning_rate, learning_rate_blob=None, policy="fixed", iter_val=0, kwargs)
def	add_lr_multiplier (self, lr_multiplier)
def	get_auxiliary_parameters (self)
def	scale_learning_rate (self, args, kwargs)
def	create_lars_inputs (self, param_init_net, weight_decay, trust, lr_max)

Public Attributes
	alpha
	epsilon
	decay
	policy
	sparse_dedup_aggregator
	engine
	init_kwargs

Additional Inherited Members
Static Public Member Functions inherited from caffe2.python.optimizer.Optimizer
def	dedup (net, sparse_dedup_aggregator, grad)

Detailed Description

Definition at line 738 of file optimizer.py.

Constructor & Destructor Documentation

def caffe2.python.optimizer.AdadeltaOptimizer.__init__	(		self,
			alpha = 0.01,
			epsilon = 1e-4,
			decay = 0.95,
			policy = "fixed",
			sparse_dedup_aggregator = None,
			engine = '',
			kwargs
	)

Constructor function to add Adadelta Optimizer

Args:
    alpha: learning rate
    epsilon: attribute of Adadelta to avoid numerical issues
    decay: attribute of Adadelta to decay the squared gradient sum
    policy: specifies how learning rate should be applied, options are
      "fixed", "step", "exp", etc.
    sparse_dedup_aggregator: specifies deduplication strategy for
      gradient slices. Works while using sparse gradients. Options
      include "mean" and "sum".
    engine: the engine used, options include "", "CUDNN", etc.

Definition at line 740 of file optimizer.py.

---

The documentation for this class was generated from the following file:

• caffe2/python/optimizer.py