def test(self):
#output from custom implementation
X = torch.Tensor([[1, 1, 1, 0, 0], [0, 1, 1, 1, 0], [0, 0, 1, 1, 1],
[0, 0, 1, 1, 0], [0, 1, 1, 0, 0]])
X = X.reshape(
(1, 1, 5, 5)) # (batch_size x channel_size x height x weight)
conv = Conv()
K = torch.Tensor([[[[1, 0, 1], [0, 1, 0], [1, 0, 1]]]])
conv.init_params(K, 3, stride=1, padding=1)
output_custom = conv.forward(X)
print(output_custom)
#ouput from pytorch's implementation
output_pytorch = F.conv2d(X, K, padding=1, stride=1)
print(output_pytorch)
self.assertEqual(output_custom, output_pytorch)
num_labels = 26
cuda = torch.cuda.is_available()
print("cuda : ")
print(cuda)
# Instantiate the CRF model
crf = CRF(input_dim, embed_dim, conv_shapes, num_labels, batch_size)
# Setup the optimizer
# opt = optim.LBFGS(crf.parameters())
opt = optim.Adam(crf.parameters())
from conv import Conv
convLayer = Conv()
convLayer.init_params(kernel_size=5)
##################################################
# Begin training
##################################################
step = 0
# Fetch dataset
dataset = get_dataset()
# split = int(0.5 * len(dataset.data)) # train-test split
split = int(0.01 * len(dataset.data)) # train-test split
# train_data, test_data = dataset.data[:split], dataset.data[split:]
# train_target, test_target = dataset.target[:split], dataset.target[split:]
train_data = dataset.data[:split]
test_data = train_data
train_target = dataset.target[:split]