调用自定义类
概述
在静态图模式下,通过使用jit_class
修饰自定义类,用户可以创建、调用该自定义类的实例,并且可以获取其属性和方法。
jit_class
应用于静态图模式,扩充完善静态图编译语法的支持范围。在动态图模式即PyNative模式下,jit_class
的使用不影响PyNative模式的执行逻辑。
本文档主要介绍jit_class
装饰器的使用方法和使用须知,以便您可以更有效地使用jit_class
装饰器功能。
jit_class修饰自定义类
使用@jit_class修饰自定义类后,支持创建、调用该自定义类的实例,获取其属性和方法。
[1]:
import numpy as np
import mindspore.nn as nn
import mindspore as ms
@ms.jit_class
class InnerNet:
value = ms.Tensor(np.array([1, 2, 3]))
class Net(nn.Cell):
def construct(self):
return InnerNet().value
ms.set_context(mode=ms.GRAPH_MODE)
net = Net()
out = net()
print(out)
[1 2 3]
jit_class支持自定义类嵌套使用、自定义类与nn.Cell嵌套使用的场景。需要注意的是,类继承时,如果父类使用了jit_class,子类也会具有jit_class的能力。
[2]:
import numpy as np
import mindspore.nn as nn
import mindspore as ms
@ms.jit_class
class Inner:
def __init__(self):
self.value = ms.Tensor(np.array([1, 2, 3]))
@ms.jit_class
class InnerNet:
def __init__(self):
self.inner = Inner()
class Net(nn.Cell):
def __init__(self):
super(Net, self).__init__()
self.inner_net = InnerNet()
def construct(self):
out = self.inner_net.inner.value
return out
ms.set_context(mode=ms.GRAPH_MODE)
net = Net()
out = net()
print(out)
[1 2 3]
jit_class仅支持修饰自定义类,不支持nn.Cell和非class类型。执行下面用例,将会出现报错。
import mindspore.nn as nn
import mindspore as ms
@ms.jit_class
class Net(nn.Cell):
def construct(self, x):
return x
ms.set_context(mode=ms.GRAPH_MODE)
x = ms.Tensor(1)
net = Net()
net(x)
报错信息如下:
TypeError: Decorator jit_class is used for user-defined classes and cannot be used for nn.Cell: Net<>.
import mindspore as ms
@ms.jit_class
def func(x, y):
return x + y
func(1, 2)
报错信息如下:
TypeError: Decorator jit_class can only be used for class type, but got <function func at 0x7fee33c005f0>.
获取自定义类的属性和方法
支持通过类名调用类的属性,不支持通过类名调用类的方法。对于类的实例,支持调用其属性和方法。
[3]:
import mindspore.nn as nn
import mindspore as ms
@ms.jit_class
class InnerNet:
def __init__(self, val):
self.number = val
def act(self, x, y):
return self.number * (x + y)
class Net(nn.Cell):
def __init__(self):
super(Net, self).__init__()
self.inner_net = InnerNet(2)
def construct(self, x, y):
return self.inner_net.number + self.inner_net.act(x, y)
ms.set_context(mode=ms.GRAPH_MODE)
x = ms.Tensor(2, dtype=ms.int32)
y = ms.Tensor(3, dtype=ms.int32)
net = Net()
out = net(x, y)
print(out)
12
不支持调用私有属性和魔术方法,所调用的方法函数须在静态图编译支持的语法范围内。执行下面用例,将会出现报错。
import numpy as np
import mindspore.nn as nn
import mindspore as ms
@ms.jit_class
class InnerNet:
def __init__(self):
self.value = ms.Tensor(np.array([1, 2, 3]))
class Net(nn.Cell):
def __init__(self):
super(Net, self).__init__()
self.inner_net = InnerNet()
def construct(self):
out = self.inner_net.__str__()
return out
ms.set_context(mode=ms.GRAPH_MODE)
net = Net()
out = net()
报错信息如下:
RuntimeError: `__str__` is a private variable or magic method, which is not supported.
创建自定义类的实例
静态图模式下,创建自定义类的实例时,参数要求为常量。
[4]:
import numpy as np
import mindspore.nn as nn
import mindspore as ms
@ms.jit_class
class InnerNet:
def __init__(self, val):
self.number = val + 3
class Net(nn.Cell):
def construct(self):
net = InnerNet(2)
return net.number
ms.set_context(mode=ms.GRAPH_MODE)
net = Net()
out = net()
print(out)
5
对于其它场景,创建自定义类的实例时,没有参数必须是常量的限制。例如下面的用例:
[5]:
import numpy as np
import mindspore.nn as nn
import mindspore as ms
@ms.jit_class
class InnerNet:
def __init__(self, val):
self.number = val + 3
class Net(nn.Cell):
def __init__(self, val):
super(Net, self).__init__()
self.inner = InnerNet(val)
def construct(self):
return self.inner.number
ms.set_context(mode=ms.GRAPH_MODE)
x = ms.Tensor(2, dtype=ms.int32)
net = Net(x)
out = net()
print(out)
5
调用自定义类的实例
调用自定义类的实例时,将会调用该类的__call__
函数方法。
[6]:
import numpy as np
import mindspore.nn as nn
import mindspore as ms
@ms.jit_class
class InnerNet:
def __init__(self, number):
self.number = number
def __call__(self, x, y):
return self.number * (x + y)
class Net(nn.Cell):
def construct(self, x, y):
net = InnerNet(2)
out = net(x, y)
return out
ms.set_context(mode=ms.GRAPH_MODE)
x = ms.Tensor(2, dtype=ms.int32)
y = ms.Tensor(3, dtype=ms.int32)
net = Net()
out = net(x, y)
print(out)
10
如果该类没有定义__call__
函数,将会报错提示。执行下面用例,将会出现报错。
import numpy as np
import mindspore.nn as nn
import mindspore as ms
@ms.jit_class
class InnerNet:
def __init__(self, number):
self.number = number
class Net(nn.Cell):
def construct(self, x, y):
net = InnerNet(2)
out = net(x, y)
return out
ms.set_context(mode=ms.GRAPH_MODE)
x = ms.Tensor(2, dtype=ms.int32)
y = ms.Tensor(3, dtype=ms.int32)
net = Net()
out = net(x, y)
print(out)
报错信息如下:
RumtimeError: MsClassObject: 'InnerNet' has no `__call__` function, please check the code.