/
linear.cr
53 lines (47 loc) · 2 KB
/
linear.cr
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
# Copyright (c) 2020 Crystal Data Contributors
#
# MIT License
#
# Permission is hereby granted, free of charge, to any person obtaining
# a copy of this software and associated documentation files (the
# "Software"), to deal in the Software without restriction, including
# without limitation the rights to use, copy, modify, merge, publish,
# distribute, sublicense, and/or sell copies of the Software, and to
# permit persons to whom the Software is furnished to do so, subject to
# the following conditions:
#
# The above copyright notice and this permission notice shall be
# included in all copies or substantial portions of the Software.
#
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
# EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
# MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
# NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
# LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
# OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
# WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
class Num::NN::LinearLayer(T) < Num::NN::Layer(T)
getter weights : Num::Grad::Variable(T)
getter bias : Num::Grad::Variable(T)
def initialize(context : Num::Grad::Context(T), inp_dim : Int, outp_dim : Int)
w = Num::NN.kaiming_normal(outp_dim, inp_dim, dtype: T)
b = T.zeros([1, outp_dim])
@weights = context.variable(w)
@bias = context.variable(b)
end
def forward(input : Num::Grad::Variable(T)) : Num::Grad::Variable(T)
output = input.value.matmul(@weights.value.transpose) + bias.value
result = input.context.variable(output)
if input.is_grad_needed || @weights.is_grad_needed || @bias.is_grad_needed
gate = Num::NN::LinearGate.new(input, @weights, @bias)
gate.cache(result, input, @weights, @bias)
end
result
end
def variables : Array(Num::Grad::Variable(T))
[weights, bias]
end
def output_shape : Array(Int32)
[@weights.value.shape[0]]
end
end