def __init__(self, params, learning_rate=1e-3, beta1=0.9, beta2=0.999, eps=1e-8, use_locking=False,
use_nesterov=False, weight_decay=0.0, loss_scale=1.0):
super(LazyAdam, self).__init__(learning_rate, params, weight_decay, loss_scale)
_check_param_value(beta1, beta2, eps, weight_decay, self.cls_name)
validator.check_value_type("use_locking", use_locking, [bool], self.cls_name)
validator.check_value_type("use_nesterov", use_nesterov, [bool], self.cls_name)
self.beta1 = Tensor(beta1, mstype.float32)
self.beta2 = Tensor(beta2, mstype.float32)
self.beta1_power = Parameter(initializer(1, [1], mstype.float32), name="beta1_power")
self.beta2_power = Parameter(initializer(1, [1], mstype.float32), name="beta2_power")
self.eps = Tensor(eps, mstype.float32)
self.use_nesterov = use_nesterov
self.use_locking = use_locking
self._is_device = True
self.moment1 = self.parameters.clone(prefix="moment1", init='zeros')
self.moment2 = self.parameters.clone(prefix="moment2", init='zeros')
self.hyper_map = C.HyperMap()
self.opt = P.Adam(use_locking, use_nesterov)
self.sparse_opt = P.FusedSparseLazyAdam(use_locking, use_nesterov)
self.sparse_opt.add_prim_attr("primitive_target", "CPU")
self._ps_pull = P.Pull()
self._ps_push = P.Push("Adam", [0, 1, 2])
self._ps_push.add_prim_attr("use_nesterov", use_nesterov)
def __init__(self,
params,
learning_rate=1e-3,
beta1=0.9,
beta2=0.999,
eps=1e-8,
use_locking=False,
use_nesterov=False,
weight_decay=0.0,
loss_scale=1.0):
super(Adam, self).__init__(learning_rate, params, weight_decay,
loss_scale)
_check_param_value(beta1, beta2, eps, weight_decay, self.cls_name)
validator.check_value_type("use_locking", use_locking, [bool],
self.cls_name)
validator.check_value_type("use_nesterov", use_nesterov, [bool],
self.cls_name)
self.beta1 = Tensor(beta1, mstype.float32)
self.beta2 = Tensor(beta2, mstype.float32)
self.beta1_power = Parameter(initializer(1, [1], mstype.float32),
name="beta1_power")
self.beta2_power = Parameter(initializer(1, [1], mstype.float32),
name="beta2_power")
self.eps = eps
self.moment1 = self.parameters.clone(prefix="moment1", init='zeros')
self.moment2 = self.parameters.clone(prefix="moment2", init='zeros')
self.hyper_map = C.HyperMap()
self.opt = P.Adam(use_locking, use_nesterov)
self.sparse_opt = P.SparseApplyAdam(use_locking, use_nesterov)
def __init__(self, params, learning_rate=1e-3, beta1=0.9, beta2=0.999, eps=1e-8, use_locking=False,
use_nesterov=False, weight_decay=0.0, loss_scale=1.0):
super(Adam, self).__init__(learning_rate, params, weight_decay, loss_scale)
_check_param_value(beta1, beta2, eps, weight_decay, self.cls_name)
validator.check_value_type("use_locking", use_locking, [bool], self.cls_name)
validator.check_value_type("use_nesterov", use_nesterov, [bool], self.cls_name)
validator.check_value_type("loss_scale", loss_scale, [float], self.cls_name)
validator.check_number_range("loss_scale", loss_scale, 1.0, float("inf"), Rel.INC_LEFT, self.cls_name)
self.beta1 = Tensor(beta1, mstype.float32)
self.beta2 = Tensor(beta2, mstype.float32)
self.beta1_power = Parameter(initializer(1, [1], mstype.float32), name="beta1_power")
self.beta2_power = Parameter(initializer(1, [1], mstype.float32), name="beta2_power")
self.eps = eps
self.moment1 = self.parameters.clone(prefix="moment1", init='zeros')
self.moment2 = self.parameters.clone(prefix="moment2", init='zeros')
self.hyper_map = C.HyperMap()
self.opt = P.Adam(use_locking, use_nesterov)
self.pow = P.Pow()
self.sqrt = P.Sqrt()
self.one = Tensor(np.array([1.0]).astype(np.float32))
self.realdiv = P.RealDiv()
def __init__(self,
params,
learning_rate=1e-3,
beta1=0.9,
beta2=0.999,
eps=1e-8,
use_locking=False,
use_nesterov=False,
weight_decay=0.0,
loss_scale=1.0,
decay_filter=lambda x: 'beta' not in x.name and 'gamma' not in
x.name):
super(Adam, self).__init__(learning_rate, params)
_check_param_value(beta1, beta2, eps, weight_decay)
validator.check_type("use_locking", use_locking, [bool])
validator.check_type("use_nesterov", use_nesterov, [bool])
validator.check_type("loss_scale", loss_scale, [float])
validator.check_number_range("loss_scale", loss_scale, 1.0,
float("inf"), Rel.INC_LEFT)
self.dynamic_lr = False
if isinstance(learning_rate, Iterable) or \
(isinstance(learning_rate, Tensor) and learning_rate.dim() == 1):
self.dynamic_lr = True
self.gather = P.GatherV2()
self.assignadd = P.AssignAdd()
self.global_step = Parameter(initializer(0, [1], mstype.int32),
name="global_step")
self.axis = 0
self.beta1 = Tensor(beta1, mstype.float32)
self.beta2 = Tensor(beta2, mstype.float32)
self.beta1_power = Parameter(initializer(1, [1], mstype.float32),
name="beta1_power")
self.beta2_power = Parameter(initializer(1, [1], mstype.float32),
name="beta2_power")
self.eps = eps
self.moment1 = self.parameters.clone(prefix="moment1", init='zeros')
self.moment2 = self.parameters.clone(prefix="moment2", init='zeros')
self.decay_tf = tuple(decay_filter(x) for x in self.parameters)
self.hyper_map = C.HyperMap()
self.opt = P.Adam(use_locking, use_nesterov)
self.weight_decay = weight_decay * loss_scale
self.reciprocal_scale = 1.0 / loss_scale
self.pow = P.Pow()
self.sqrt = P.Sqrt()
self.one = Tensor(np.array([1.0]).astype(np.float32))
self.realdiv = P.RealDiv()
def __init__(self, var, m, v):
super(AdamNet, self).__init__()
self.apply_adam = P.Adam()
self.var = Parameter(var, name="var")
self.m = Parameter(m, name="m")
self.v = Parameter(v, name="v")