pytorch masked fill方法簡單理解

>>

>
help
(torch.tensor.masked_fill)
help on method_descriptor:
masked_fill(..
.)masked_fill(mask, value)
-> tensor
out-of-place version of :meth:`torch.tensor.masked_fill_`
>>
>
help
(torch.tensor.masked_fill_)
help on method_descriptor:
masked_fill_(..
.)masked_fill_(mask, value)
fills elements of :attr:`self` tensor with
:attr:`value` where :attr:`mask` is
true
. the shape of :attr:`mask` must be
:ref:`broadcastable 
` with the shape of the underlying
tensor.
args:
mask (booltensor)
: the boolean mask
value (
float
): the value to fill in
with

>>

> t = torch.randn(3,
2)>>
> t
tensor([[
-0.9180,-
0.4654],
[0.9866,-
1.3063],
[1.8359
,1.1607]]
)>>
> m = torch.randint(0,
2,(3
,2))
>>
> m
tensor([[
0,1]
,[1,
1],[
1,0]
])>>
> m ==
0tensor([[
true
,false],
[false
,false],
[false
,true]]
)>>
> t.masked_fill(m ==0,
-1e9
)tensor([[
-1.0000e+09,-
4.6544e-01],
[9.8660e-01,-
1.3063e+00],
[1.8359e+00,-
1.0000e+09]]
)

>>

> m = torch.randint(0,
2,(3
,1))
>>
> m
tensor([[
0],[
1],[
1]])
# m和t是可廣播的
>>
> t.masked_fill(m ==0,
-1e9
)tensor([[
-1.0000e+09,-
1.0000e+09],
[9.8660e-01,-
1.3063e+00],
[1.8359e+00
,1.1607e+00]]
)>>
> m = torch.randint(0,
2,(3
,3))
>>
> m
tensor([[
0,0,
0],[
1,1,
0],[
1,1,
1]])
# m和t是不可廣播的
>>
> t.masked_fill(m ==0,
-1e9
)traceback (most recent call last)
:  file ""
, line 1,in
runtimeerror: the size of tensor a (
3) must match the size of tensor b (
2) at non-singleton dimension 1

pytorch masked fill方法簡單理解

help torch.tensor.masked fill help on method descriptor masked fill masked fill mask,value tensor out of place version of meth torch.tensor.masked fil...

Spring Cloud Eureka簡單理解

eureka服務治理基礎架構包括三個核心要素。1 服務註冊中心 eureka分為客戶端和服務端，eureka服務端提供服務註冊與發現的功能。2 服務提供者 提供服務的應用，spring boot應用或者遵循eureka通訊機制的應用。將應用自己註冊到eureka註冊中心，以供其它應用的發現。3 服務...

Java Method的invoke簡單理解

首先method類代表乙個方法，所以invoke 呼叫 就是呼叫method類代表的方法。它可以讓你實現動態呼叫，例如你可以動態的傳人引數。下面是乙個簡單的例子。public class methodtest class clazz test.class trycatch nosuchmethode...