!709 fix weight decay error in optimizer AdamWeightDecay

Merge pull request !709 from wangnan39/fix_bug_in_adamweightdecay

mindspore/nn/optim/adam.py

@@ -31,8 +31,8 @@ _learning_rate_update_func = ['linear', 'cos', 'sin']
 adam_opt = C.MultitypeFuncGraph("adam_opt")
 
 
-@adam_opt.register("Tensor", "Tensor", "Tensor", "Tensor", "Tensor", "Tensor", "Tensor", "Tensor", "Tensor")
-def _update_run_op(beta1, beta2, eps, lr, weight_decay_tensor, param, m, v, gradient):
+@adam_opt.register("Tensor", "Tensor", "Tensor", "Tensor", "Tensor", "Tensor", "Tensor", "Tensor", "Tensor", "Bool")
+def _update_run_op(beta1, beta2, eps, lr, weight_decay_tensor, param, m, v, gradient, decay_flag):
     """
     Update parameters.
 
@@ -67,7 +67,8 @@ def _update_run_op(beta1, beta2, eps, lr, weight_decay_tensor, param, m, v, grad
     next_v = op_mul(beta2, v) + op_mul(op_cast(F.tuple_to_array((1.0,)), mstype.float32) - beta2, op_square(gradient))
 
     update = next_m / (op_sqrt(next_v) + eps)
-    update = update + op_mul(weight_decay_tensor, param)
+    if decay_flag:
+        update = update + op_mul(weight_decay_tensor, param)
 
     update_with_lr = op_mul(lr, update)
     next_param = param - op_reshape(update_with_lr, op_shape(param))
@@ -90,6 +91,17 @@ def _check_param_value(beta1, beta2, eps, weight_decay, prim_name):
     validator.check_number_range("weight_decay", weight_decay, 0.0, float("inf"), Rel.INC_LEFT, prim_name)
 
 
+def _check_learning_rate_value(learning_rate, end_learning_rate, decay_steps, power, prim_name):
+    """Check the type of inputs."""
+    validator.check_float_positive('learning_rate', learning_rate, prim_name)
+    validator.check_float_legal_value('learning_rate', learning_rate, prim_name)
+    validator.check_float_positive('end_learning_rate', end_learning_rate, prim_name)
+    validator.check_float_legal_value('end_learning_rate', end_learning_rate, prim_name)
+    validator.check_float_positive('power', power, prim_name)
+    validator.check_float_legal_value('power', power, prim_name)
+    validator.check_integer('decay_steps', decay_steps, 0, Rel.GT, prim_name)
+
+
 @adam_opt.register("Function", "Tensor", "Tensor", "Tensor", "Tensor", "Tensor", "Number", "Tensor", "Tensor", "Tensor",
                    "Tensor")
 def _run_opt_with_one_number(opt, lr, beta1_power, beta2_power, beta1, beta2, eps, gradient, params, moment1,
@@ -126,8 +138,13 @@ class Adam(Optimizer):
     Args:
         params (list[Parameter]): A list of parameter, which will be updated. The element in `params`
                                   should be class mindspore.Parameter.
-        learning_rate (Union[float, Tensor, Iterable]): The Learning rate.
-            Iterable type is used for the dynamic learning rate.
+        learning_rate (Union[float, Tensor, Iterable]): A value for the learning rate. When the learning_rate is
+                                                        Iterable or a Tensor and the dims of the Tensor is 1,
+                                                        use dynamic learning rate, then the i-th step will
+                                                        take the i-th value as the learning rate.
+                                                        When the learning_rate is float or learning_rate is a Tensor
+                                                        but the dims of the Tensor is 0, use fixed learning rate.
+                                                        Other cases are not supported. Default: 1e-3.
         beta1 (float): The exponential decay rate for the 1st moment estimates. Should be in range (0.0, 1.0).
         beta2 (float): The exponential decay rate for the 2nd moment estimates. Should be in range (0.0, 1.0).
         eps (float): Term added to the denominator to improve numerical stability. Should be greater than 0.
@@ -140,6 +157,8 @@ class Adam(Optimizer):
         weight_decay (float): Weight decay (L2 penalty). Default: 0.0.
         loss_scale (float): A floating point value for the loss scale. Default: 1.0.
             Should be equal to or greater than 1.
+        decay_filter (Function): A function to determine whether to apply weight decay on parameters. Default:
+                                 lambda x: 'LayerNorm' not in x.name and 'bias' not in x.name.
 
     Inputs:
         - **gradients** (tuple[Tensor]) - The gradients of `params`, the shape is the same as `params`.
@@ -207,7 +226,13 @@ class AdamWeightDecay(Optimizer):
     Args:
         params (list[Parameter]): A list of parameter, which will be updated. The element in `params`
                                   should be class mindspore.Parameter.
-        learning_rate (float): A floating point value for the learning rate. Default: 1e-3.
+        learning_rate (Union[float, Tensor, Iterable]): A value for the learning rate. When the learning_rate is
+                                                        Iterable or a Tensor and the dims of the Tensor is 1,
+                                                        use dynamic learning rate, then the i-th step will
+                                                        take the i-th value as the learning rate.
+                                                        When the learning_rate is float or learning_rate is a Tensor
+                                                        but the dims of the Tensor is 0, use fixed learning rate.
+                                                        Other cases are not supported. Default: 1e-3.
         beta1 (float): The exponential decay rate for the 1st moment estimates. Default: 0.9.
             Should be in range (0.0, 1.0).
         beta2 (float): The exponential decay rate for the 2nd moment estimates. Default: 0.999.
@@ -215,6 +240,8 @@ class AdamWeightDecay(Optimizer):
         eps (float): Term added to the denominator to improve numerical stability. Default: 1e-6.
             Should be greater than 0.
         weight_decay (float): Weight decay (L2 penalty). Default: 0.0.
+        decay_filter (Function): A function to determine whether to apply weight decay on parameters. Default:
+                                 lambda x: 'LayerNorm' not in x.name and 'bias' not in x.name.
 
     Inputs:
         - **gradients** (tuple[Tensor]) - The gradients of `params`, the shape is the same as `params`.
@@ -228,10 +255,10 @@ class AdamWeightDecay(Optimizer):
         >>> optim = nn.AdamWeightDecay(params=net.trainable_params())
         >>> model = Model(net, loss_fn=loss, optimizer=optim, metrics=None)
    """
-    def __init__(self, params, learning_rate=1e-3, beta1=0.9, beta2=0.999, eps=1e-6, weight_decay=0.0):
+    def __init__(self, params, learning_rate=1e-3, beta1=0.9, beta2=0.999, eps=1e-6, weight_decay=0.0,
+                 decay_filter=lambda x: 'beta' not in x.name and 'gamma' not in x.name):
         super(AdamWeightDecay, self).__init__(learning_rate, params)
         _check_param_value(beta1, beta2, eps, weight_decay, self.cls_name)
-        self.lr = Tensor(np.array([learning_rate]).astype(np.float32))
         self.beta1 = Tensor(np.array([beta1]).astype(np.float32))
         self.beta2 = Tensor(np.array([beta2]).astype(np.float32))
         self.eps = Tensor(np.array([eps]).astype(np.float32))
@@ -240,13 +267,15 @@ class AdamWeightDecay(Optimizer):
         self.params = self.parameters
         self.moments1 = self.params.clone(prefix="adam_m", init='zeros')
         self.moments2 = self.params.clone(prefix="adam_v", init='zeros')
+        self.decay_flag = tuple(decay_filter(x) for x in self.params)
 
         self.hyper_map = C.HyperMap()
 
     def construct(self, gradients):
-        updated_velocity = self.hyper_map(F.partial(adam_opt, self.beta1, self.beta2, self.eps, self.lr,
+        lr = self.get_lr()
+        updated_velocity = self.hyper_map(F.partial(adam_opt, self.beta1, self.beta2, self.eps, lr,
                                                     self.weight_decay_tensor),
-                                          self.params, self.moments1, self.moments2, gradients)
+                                          self.params, self.moments1, self.moments2, gradients, self.decay_flag)
 
         return updated_velocity
 
@@ -269,6 +298,8 @@ class AdamWeightDecayDynamicLR(Optimizer):
         eps (float): Term added to the denominator to improve numerical stability. Default: 1e-6.
             Should be greater than 0.
         weight_decay (float): Weight decay (L2 penalty). Default: 0.0.
+        decay_filter (Function): A function to determine whether to apply weight decay on parameters. Default:
+                                 lambda x: 'LayerNorm' not in x.name and 'bias' not in x.name.
 
     Inputs:
         - **gradients** (tuple[Tensor]) - The gradients of `params`, the shape is the same as `params`.
@@ -291,10 +322,11 @@ class AdamWeightDecayDynamicLR(Optimizer):
                  beta1=0.9,
                  beta2=0.999,
                  eps=1e-6,
-                 weight_decay=0.0):
+                 weight_decay=0.0,
+                 decay_filter=lambda x: 'beta' not in x.name and 'gamma' not in x.name):
         super(AdamWeightDecayDynamicLR, self).__init__(learning_rate, params)
         _check_param_value(beta1, beta2, eps, weight_decay, self.cls_name)
-
+        _check_learning_rate_value(learning_rate, end_learning_rate, decay_steps, power, self.cls_name)
         # turn them to scalar when me support scalar/tensor mix operations
         self.global_step = Parameter(initializer(0, [1]), name="global_step")
         self.decay_steps = Tensor(np.array([decay_steps]).astype(np.float32))
@@ -308,7 +340,7 @@ class AdamWeightDecayDynamicLR(Optimizer):
         self.params = self.parameters
         self.moments1 = self.params.clone(prefix="adam_m", init='zeros')
         self.moments2 = self.params.clone(prefix="adam_v", init='zeros')
-
+        self.decay_flag = tuple(decay_filter(x) for x in self.params)
         self.hyper_map = C.HyperMap()
         self.min = P.Minimum()
         self.pow = P.Pow()
@@ -320,7 +352,7 @@ class AdamWeightDecayDynamicLR(Optimizer):
         lr = self.diff_learning_rate * self.pow(self.one - p, self.power) + self.end_learning_rate
         updated_velocity = self.hyper_map(F.partial(adam_opt, self.beta1, self.beta2, self.eps, lr,
                                                     self.weight_decay_tensor),
-                                          self.params, self.moments1, self.moments2, gradients)
+                                          self.params, self.moments1, self.moments2, gradients, self.decay_flag)
 
         added_global_step = self.global_step + self.one
         F.control_depend(lr, added_global_step)

mindspore/nn/optim/lamb.py

@@ -112,16 +112,18 @@ def _check_param_value(decay_steps, warmup_steps, start_learning_rate,
                        end_learning_rate, power, beta1, beta2, eps, weight_decay, prim_name):
 
     """Check the type of inputs."""
-    validator.check_value_type("decay_steps", decay_steps, [int], prim_name)
-    validator.check_value_type("warmup_steps", warmup_steps, [int], prim_name)
-    validator.check_value_type("start_learning_rate", start_learning_rate, [float], prim_name)
-    validator.check_value_type("end_learning_rate", end_learning_rate, [float], prim_name)
-    validator.check_value_type("power", power, [float], prim_name)
+    validator.check_float_positive('start_learning_rate', start_learning_rate, prim_name)
+    validator.check_float_legal_value('start_learning_rate', start_learning_rate, prim_name)
+    validator.check_float_positive('end_learning_rate', end_learning_rate, prim_name)
+    validator.check_float_legal_value('end_learning_rate', end_learning_rate, prim_name)
+    validator.check_float_positive('power', power, prim_name)
+    validator.check_float_legal_value('power', power, prim_name)
+    validator.check_integer('decay_steps', decay_steps, 0, Rel.GT, prim_name)
+    validator.check_integer('warmup_steps', decay_steps, 0, Rel.GT, prim_name)
     validator.check_value_type("beta1", beta1, [float], prim_name)
     validator.check_value_type("beta2", beta2, [float], prim_name)
     validator.check_value_type("eps", eps, [float], prim_name)
     validator.check_value_type("weight_dacay", weight_decay, [float], prim_name)
-    validator.check_number_range("decay_steps", decay_steps, 1, float("inf"), Rel.INC_LEFT, prim_name)
     validator.check_number_range("beta1", beta1, 0.0, 1.0, Rel.INC_NEITHER, prim_name)
     validator.check_number_range("beta2", beta2, 0.0, 1.0, Rel.INC_NEITHER, prim_name)
     validator.check_number_range("eps", eps, 0.0, float("inf"), Rel.INC_NEITHER, prim_name)

mindspore/nn/optim/sgd.py

@@ -42,7 +42,13 @@ class SGD(Optimizer):
     Args:
         params (list[Parameter]): A list of parameter, which will be updated. The element in `params`
                                   should be class mindspore.Parameter.
-        learning_rate (float): A floating point value for the learning rate. Default: 0.1.
+        learning_rate (Union[float, Tensor, Iterable]): A value for the learning rate. When the learning_rate is
+                                                        Iterable or a Tensor and the dims of the Tensor is 1,
+                                                        use dynamic learning rate, then the i-th step will
+                                                        take the i-th value as the learning rate.
+                                                        When the learning_rate is float or learning_rate is a Tensor
+                                                        but the dims of the Tensor is 0, use fixed learning rate.
+                                                        Other cases are not supported. Default: 0.1.
         momentum (float): A floating point value the momentum. Default: 0.
         dampening (float): A floating point value of dampening for momentum. Default: 0.
         weight_decay (float): Weight decay (L2 penalty). Default: 0.