graphtorch's Introduction

graphtorch

Package converts sparse graph matrix to PyTorch model

Installation

pip install graphtorch

Examples

Create sparse matrix with essential impormation

from graphtorch import SpraseMatrix

mat1 = np.array([[0,2,0,0,2,0,0,0,0,0],
                [2,0,2,0,0,0,0,0,0,0],
                [0,2,0,2,0,0,0,0,0,0],
                [0,0,0,0,0,1,1,0,0,0],
                [0,0,0,0,0,0,1,1,0,0],
                [0,0,0,0,0,0,0,0,0,3],
                [0,0,0,0,0,0,0,0,3,0]])  
in_dim = 5   
out_dim = 2  
mat_wann1 = SparseMatrix(mat1, in_dim, out_dim)

Create sparse torch model with SparseMatrix

from graphtorch import SparseModel

activations = [None, None, nn.ReLU(), nn.Sigmoid()]  
constant_weight = 1 
model = SparseModel(mat_wann1, activations, constant_weight)

numpy_input = np.array([[1,2,3,4,5],  
                        [6,7,8,9,10],  
                        [11,12,13,14,15]])      

numpy_input = torch.from_numpy(numpy_input).float()  
output, nodes = model(numpy_input)

output

tensor([[1.0000, 1.0000],
        [1.0000, 1.0000],
        [1.0000, 1.0000]], grad_fn=<CatBackward>)

nodes

{'hidden_0': tensor([[ 7.],
         [17.],
         [27.]], grad_fn=<AddBackward0>), 'hidden_1': tensor([[ 4.],
         [14.],
         [24.]], grad_fn=<AddBackward0>), 'hidden_2': tensor([[ 6.],
         [16.],
         [26.]], grad_fn=<AddBackward0>), 'hidden_3': tensor([[11.],
         [31.],
         [51.]], grad_fn=<AddBackward0>), 'hidden_4': tensor([[10.],
         [30.],
         [50.]], grad_fn=<AddBackward0>), 'output_0': tensor([[1.0000],
         [1.0000],
         [1.0000]], grad_fn=<SigmoidBackward>), 'output_1': tensor([[1.0000],
         [1.0000],
         [1.0000]], grad_fn=<SigmoidBackward>)}