def test_rbm_pseudo_likelihood():
new_rbm = rbm.RBM()
samples = torch.ones(1, 128)
pl = new_rbm.pseudo_likelihood(samples)
assert pl.detach().numpy() < 0
def test_rbm_energy():
new_rbm = rbm.RBM()
samples = torch.ones(1, 128)
energy = new_rbm.energy(samples)
assert energy.detach().numpy() < 0
def test_rbm_forward():
new_rbm = rbm.RBM()
v = torch.ones(1, 128)
probs = new_rbm.forward(v)
assert probs.size(1) == 128
def test_rbm_b_setter():
new_rbm = rbm.RBM()
try:
new_rbm.b = 1
except:
new_rbm.b = torch.nn.Parameter(torch.zeros(128))
assert new_rbm.b.size(0) == 128
def test_rbm_optimizer_setter():
new_rbm = rbm.RBM()
try:
new_rbm.optimizer = 'OPT'
except:
new_rbm.optimizer = torch.optim.SGD(new_rbm.parameters(), lr=0.1)
assert type(new_rbm.optimizer).__name__ == 'SGD'
def test_rbm_visible_sampling():
new_rbm = rbm.RBM()
h = torch.ones(1, 128)
probs, states = new_rbm.visible_sampling(h)
assert probs.size(1) == 128
assert states.size(1) == 128
def test_rbm_hidden_sampling():
new_rbm = rbm.RBM()
v = torch.ones(1, 128)
probs, states = new_rbm.hidden_sampling(v)
assert probs.size(1) == 128
assert states.size(1) == 128
def test_rbm_W_setter():
new_rbm = rbm.RBM()
try:
new_rbm.W = 1
except:
new_rbm.W = torch.nn.Parameter(torch.randn(128, 128) * 0.01)
assert new_rbm.W.size(0) == 128
assert new_rbm.W.size(1) == 128
def test_rbm_T_setter():
new_rbm = rbm.RBM()
try:
new_rbm.T = "a"
except:
new_rbm.T = 0.1
assert new_rbm.T == 0.1
try:
new_rbm.T = -1
except:
new_rbm.T = 0.1
assert new_rbm.T == 0.1
def test_rbm_decay_setter():
new_rbm = rbm.RBM()
try:
new_rbm.decay = "a"
except:
new_rbm.decay = 0.1
assert new_rbm.decay == 0.1
try:
new_rbm.decay = -1
except:
new_rbm.decay = 0.1
assert new_rbm.decay == 0.1
def test_rbm_lr_setter():
new_rbm = rbm.RBM()
try:
new_rbm.lr = "a"
except:
new_rbm.lr = 0.1
assert new_rbm.lr == 0.1
try:
new_rbm.lr = -1
except:
new_rbm.lr = 0.1
assert new_rbm.lr == 0.1
def test_rbm_steps_setter():
new_rbm = rbm.RBM()
try:
new_rbm.steps = "a"
except:
new_rbm.steps = 1
assert new_rbm.steps == 1
try:
new_rbm.steps = 0
except:
new_rbm.steps = 1
assert new_rbm.steps == 1
def test_rbm_n_visible_setter():
new_rbm = rbm.RBM()
try:
new_rbm.n_visible = "a"
except:
new_rbm.n_visible = 1
assert new_rbm.n_visible == 1
try:
new_rbm.n_visible = 0
except:
new_rbm.n_visible = 1
assert new_rbm.n_visible == 1
def test_rbm_n_hidden_setter():
new_rbm = rbm.RBM()
try:
new_rbm.n_hidden = "a"
except:
new_rbm.n_hidden = 1
assert new_rbm.n_hidden == 1
try:
new_rbm.n_hidden = 0
except:
new_rbm.n_hidden = 1
assert new_rbm.n_hidden == 1
def test_rbm_momentum_setter():
new_rbm = rbm.RBM()
try:
new_rbm.momentum = "a"
except:
new_rbm.momentum = 0.1
assert new_rbm.momentum == 0.1
try:
new_rbm.momentum = -1
except:
new_rbm.momentum = 0.1
assert new_rbm.momentum == 0.1
def test_rbm_reconstruct():
test = torchvision.datasets.KMNIST(
root='./data',
train=False,
download=True,
transform=torchvision.transforms.ToTensor())
new_rbm = rbm.RBM(n_visible=784,
n_hidden=128,
steps=1,
learning_rate=0.1,
momentum=0,
decay=0,
temperature=1,
use_gpu=False)
e, v = new_rbm.reconstruct(test)
assert e >= 0
assert v.size(1) == 784
def test_rbm_fit():
train = torchvision.datasets.KMNIST(
root='./data',
train=True,
download=True,
transform=torchvision.transforms.ToTensor())
new_rbm = rbm.RBM(n_visible=784,
n_hidden=128,
steps=1,
learning_rate=0.1,
momentum=0,
decay=0,
temperature=1,
use_gpu=False)
e, pl = new_rbm.fit(train, batch_size=128, epochs=1)
assert e >= 0
assert pl <= 0
def test_rbm_n_visible():
new_rbm = rbm.RBM()
assert new_rbm.n_visible == 128
def test_rbm_steps():
new_rbm = rbm.RBM()
assert new_rbm.steps == 1
def test_rbm_n_hidden():
new_rbm = rbm.RBM()
assert new_rbm.n_hidden == 128
def test_rbm_a():
new_rbm = rbm.RBM()
assert new_rbm.a.size(0) == 128
def test_rbm_T():
new_rbm = rbm.RBM()
assert new_rbm.T == 1
def test_rbm_decay():
new_rbm = rbm.RBM()
assert new_rbm.decay == 0
def test_rbm_optimizer_setter():
new_rbm = rbm.RBM()
new_rbm.optimizer = torch.optim.SGD(new_rbm.parameters(), lr=0.1)
assert type(new_rbm.optimizer).__name__ == "SGD"
def test_rbm_optimizer():
new_rbm = rbm.RBM()
assert type(new_rbm.optimizer).__name__ == "SGD"
def test_rbm_b():
new_rbm = rbm.RBM()
assert new_rbm.b.size(0) == 128
def test_rbm_lr():
new_rbm = rbm.RBM()
assert new_rbm.lr == 0.1
def test_rbm_momentum():
new_rbm = rbm.RBM()
assert new_rbm.momentum == 0
def test_rbm_W():
new_rbm = rbm.RBM()
assert new_rbm.W.size(0) == 128
assert new_rbm.W.size(1) == 128
