DeepJava/dot3d.java at master · dido1998/DeepJava · GitHub

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
54
55
56
57
58
59
60
61
62
63
64
65
66
packagetensordef;
importbasicops.*;

publicclassdot3dextendssuperopdef
{
tensorgraphgraph;
backpropagationstructure<dot3d> curstruct;
tensorarray3darr1;
tensorarray3darr2;
tensorarray3deval;
tensorarraysplit1[];
tensorarraysplit2[];
tensorarrayopsop;
dotdotops[];
tensorarrayout[];
publicdot3d(tensorarray3darr1,tensorarray3darr2,tensorgraphgraph)
	{
op=newtensorarrayops();
this.arr1=arr1;
this.arr2=arr2;
this.graph=graph;
if(arr1.dim1!=arr2.dim1 && arr1.dim2!=arr2.dim2 && arr1.dim3!=arr2.dim3)
 {
System.out.println("dimensions should be equal");
System.exit(1);
 }
else
 {
out=newtensorarray[arr1.dim3];
//eval=new tensorarray3d(arr1.dim1,arr1.dim2,arr1.dim3,arr1.trainable);
split1=op.convert3dto2d(arr1);
split2=op.convert3dto2d(arr2);

dotops=newdot[arr1.dim3];
for(inti=0;i<arr1.dim3;i++)
 {
dotops[i]=newdot(split1[i],split2[i],graph);
 }

//System.out.println(arr1.arr[0][0][0].data);
//System.out.println(arr2.arr[0][0][0].data);
 }
	}
publictensorarray3dforwardconv()
	{
for(inti=0;i<arr1.dim3;i++)
 {
out[i]=dotops[i].forward();
 }
eval=op.convert2dto3d(out);
curstruct=newbackpropagationstructure<dot3d>(this,null,eval);
graph.addtolist(curstruct);
returneval;
	}
publicvoidbackwardconv(tensorarray3dbackflow)
	{
//System.out.println(backflow.arr[0][0][0].grad);
for(inti=0;i<arr1.dim3;i++)
 {
dotops[i].backward(out[i]);

 }
graph.removefromlist(curstruct);

	}
}