Blocks Documentation Release 0.2.0 Université de Montréal

是 否 +2 论坛币

k人 参与回答

经管之家送您一份

应届毕业生专属福利!

求职就业群

赵安豆老师微信：zhaoandou666

经管之家联合CDA

送您一个全额奖学金名额~ !

立即领取

感谢您参与论坛问题回答

经管之家送您两个论坛币！

+2 论坛币

本帖隐藏的内容

• Blocks Documentation 0.2.0.pdf (1001.04 KB)
  2016-7-25 01:38:30 上传
• http://blocks.readthedocs.io/en/latest/
  

扫码加我 拉你入群

请注明：姓名-公司-职位

以便审核进群资格，未注明则拒绝

分享0 收藏0 回帖

关键词：Universit Document Release DOCUME Blocks latest

1. >>> import numpy
   
2. >>> import theano
   
3. >>> from theano import tensor
   
4. >>> from blocks import initialization
   
5. >>> from blocks.bricks import Identity
   
6. >>> from blocks.bricks.recurrent import SimpleRecurrent
   
7. >>> x = tensor.tensor3('x')
   
8. >>> rnn = SimpleRecurrent(
   
9. ... dim=3, activation=Identity(), weights_init=initialization.Identity())
   
10. >>> rnn.initialize()
    
11. >>> h = rnn.apply(x)
    
12. >>> f = theano.function([x], h)
    
13. >>> print(f(numpy.ones((3, 1, 3), dtype=theano.config.floatX)))
    
14. [[[ 1. 1. 1.]]
    
15. [[ 2. 2. 2.]]
    
16. [[ 3. 3. 3.]]]...

复制代码

1. >>> from blocks.bricks import Linear
   
2. >>> doubler = Linear(
   
3. ... input_dim=3, output_dim=3, weights_init=initialization.Identity(2),
   
4. ... biases_init=initialization.Constant(0))
   
5. >>> doubler.initialize()
   
6. >>> h_doubler = rnn.apply(doubler.apply(x))
   
7. >>> f = theano.function([x], h_doubler)
   
8. >>> print(f(numpy.ones((3, 1, 3), dtype=theano.config.floatX)))
   
9. [[[ 2. 2. 2.]]
   
10. [[ 4. 4. 4.]]
    
11. [[ 6. 6. 6.]]]...

复制代码

谢谢分享。

谢谢分享