def test_rsqrt():
np.random.seed(0)
x_np = np.random.rand(2, 3, 4, 4).astype(np.float32)
output_ms = P.Rsqrt()(Tensor(x_np))
output_np = 1 / np.sqrt(x_np)
assert np.allclose(output_ms.asnumpy(), output_np)
def test_sqrt():
x_np = np.random.rand(2, 3, 4, 4).astype(np.float32)
context.set_context(mode=context.PYNATIVE_MODE, device_target="GPU")
output_ms = P.Sqrt()(Tensor(x_np))
output_np = np.sqrt(x_np)
assert np.allclose(output_ms.asnumpy(), output_np)
output_ms = P.Rsqrt()(Tensor(x_np))
output_np = 1 / np.sqrt(x_np)
assert np.allclose(output_ms.asnumpy(), output_np)
def _update_run_op(beta1, beta2, eps, global_step, lr, weight_decay, param, m,
v, gradient, decay_flag, optim_filter):
"""
Update parameters.
Args:
beta1 (Tensor): The exponential decay rate for the 1st moment estimations. Should be in range (0.0, 1.0).
beta2 (Tensor): The exponential decay rate for the 2nd moment estimations. Should be in range (0.0, 1.0).
eps (Tensor): Term added to the denominator to improve numerical stability. Should be greater than 0.
lr (Tensor): Learning rate.
weight_decay (Number): Weight decay. Should be equal to or greater than 0.
global_step (Tensor): Global step.
param (Tensor): Parameters.
m (Tensor): m value of parameters.
v (Tensor): v value of parameters.
gradient (Tensor): Gradient of parameters.
decay_flag (bool): Specifies whether param update with weight decay.
optim_filter(bool): Applies parameter update or not.
Returns:
Tensor, the new value of v after updating.
"""
if optim_filter:
op_mul = P.Mul()
op_sqrt = P.Sqrt()
op_rsqrt = P.Rsqrt()
op_square = P.Square()
op_cast = P.Cast()
op_reshape = P.Reshape()
op_shape = P.Shape()
op_pow = P.Pow()
op_norm = layer.Norm()
op_select = P.Select()
op_greater = P.Greater()
op_fill = P.Fill()
op_dtype = P.DType()
param_fp32 = op_cast(param, mstype.float32)
m_fp32 = op_cast(m, mstype.float32)
v_fp32 = op_cast(v, mstype.float32)
gradient_fp32 = op_cast(gradient, mstype.float32)
next_m = op_mul(beta1, m_fp32) + op_mul(
op_cast(num_one, mstype.float32) - beta1, gradient_fp32)
next_v = op_mul(beta2, v_fp32) + op_mul(
op_cast(num_one, mstype.float32) - beta2, op_square(gradient_fp32))
next_mm = next_m / (op_cast(num_one, mstype.float32) - op_pow(
beta1, op_cast(global_step + num_one, mstype.float32)))
next_vv = next_v / (op_cast(num_one, mstype.float32) - op_pow(
beta2, op_cast(global_step + num_one, mstype.float32)))
w_norm = op_norm(param_fp32)
g_norm = op_norm(gradient_fp32)
g_norm_hat = op_norm(
op_mul(next_mm, op_rsqrt(next_vv + eps)) +
weight_decay * param_fp32)
zeros = F.zeros_like(w_norm)
ones = op_fill(op_dtype(w_norm), op_shape(w_norm), 1.0)
trust_ratio = op_select(
op_greater(w_norm, zeros),
op_select(op_greater(g_norm, zeros), w_norm / g_norm_hat, ones),
ones)
tens = op_fill(op_dtype(trust_ratio), op_shape(trust_ratio), 10.0)
trust_ratio = C.clip_by_value(trust_ratio, zeros, tens)
update = next_mm / (op_sqrt(next_vv) + eps)
if decay_flag:
update = update + op_mul(weight_decay, param_fp32)
update_with_lr = op_mul(op_mul(trust_ratio, lr), update)
next_param = param_fp32 - op_reshape(update_with_lr,
op_shape(param_fp32))
next_param = F.depend(
next_param, F.assign(param, op_cast(next_param, F.dtype(param))))
next_param = F.depend(next_param,
F.assign(m, op_cast(next_m, F.dtype(m))))
next_param = F.depend(next_param,
F.assign(v, op_cast(next_v, F.dtype(v))))
return op_cast(next_param, F.dtype(param))
return gradient
'desc_bprop': [[2, 3]]}),
('Rank', {
'block': P.Rank(),
'desc_inputs': [[2, 3]],
'skip': ['backward']}),
('InvertPermutation', {
'block': P.InvertPermutation(),
'desc_const': [(0, 3, 1, 2)],
'desc_inputs': [],
'skip': ['backward']}),
('Square', {
'block': P.Square(),
'desc_inputs': [[4]],
'desc_bprop': [[4]]}),
('Rsqrt', {
'block': P.Rsqrt(),
'desc_inputs': [[4]],
'desc_bprop': [[4]]}),
('Sqrt', {
'block': P.Sqrt(),
'desc_inputs': [[4]],
'desc_bprop': [[4]]}),
('RealDiv', {
'block': P.RealDiv(),
'desc_inputs': [[4, 5], [2, 3, 4, 5]],
'desc_bprop': [[2, 3, 4, 5]]}),
('Div', {
'block': P.Div(),
'desc_inputs': [[4, 5], [2, 3, 4, 5]],
'desc_bprop': [[2, 3, 4, 5]]}),
('Equal', {
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# ============================================================================
from mindspore.ops import Primitive
from mindspore.ops import operations as P
add = P.TensorAdd()
mul = P.Mul()
real_div = P.RealDiv()
rsqrt = P.Rsqrt()
sqrt = P.Sqrt()
make_tuple = Primitive('make_tuple')
tuple_getitem = Primitive('tuple_getitem')
LambNextMVWithDecayV1 = Primitive('LambNextMVWithDecayV1')
class FnDict:
def __init__(self):
self.fnDict = {}
def __call__(self, fn):
self.fnDict[fn.__name__] = fn
def __getitem__(self, name):
return self.fnDict[name]
def __init__(self, strategy1, strategy2):
super().__init__()
self.matmul = P.MatMul().set_strategy(strategy1)
self.rsqrt = P.Rsqrt().set_strategy(strategy2)
self.matmul2 = P.MatMul().set_strategy(strategy1)
# input two tensors, their shapes do not match
('Mul2', {
'block': (P.Mul(), {'exception': ValueError, 'error_keywords': ['Mul']}),
'desc_inputs': [Tensor(np.ones([3, 5]).astype(np.float32)), Tensor(np.ones([3, 4]).astype(np.float32))],
'skip': ['backward']}),
# input is Tensor(bool)
('Square1', {
'block': (P.Square(),
{'exception': TypeError, 'error_keywords': ['Square']}),
'desc_inputs': [Tensor(np.ones([2, 3]).astype(np.bool_))],
'skip': ['backward']}),
# input is Tensor(bool)
('Rsqrt1', {
'block': (P.Rsqrt(),
{'exception': TypeError, 'error_keywords': ['Rsqrt']}),
'desc_inputs': [Tensor(np.ones([2, 3]).astype(np.bool_))],
'skip': ['backward']}),
# input is Tensor(bool)
('Sqrt1', {
'block': (P.Sqrt(),
{'exception': TypeError, 'error_keywords': ['Sqrt']}),
'desc_inputs': [Tensor(np.ones([2, 3]).astype(np.bool_))],
'skip': ['backward']}),
# input is not Tensor
('Reciprocal1', {
'block': (P.Reciprocal(),
{'exception': TypeError, 'error_keywords': ['Reciprocal']}),
