def train_gaussianmixture():
mu_target = torch.FloatTensor([-1, 0.3])
logvar_target = torch.FloatTensor([-1, 2])
# mu_target = torch.FloatTensor([0])
# logvar_target = torch.FloatTensor([0])
target = lowd.GaussianMixture(num_mixes, z_dim, mu_target, logvar_target)
model = getattr(lowd, model_name)(target, z_dim, num_samples)
train(model)
def train_gaussianmixture():
mu_target = torch.FloatTensor([[-1, 3], [0, 3]])
logvar_target = torch.FloatTensor([[-2, 1], [-1, -2]])
# B = torch.eye(num_samples)
# mu_target = torch.FloatTensor([0])
# logvar_target = torch.FloatTensor([0])
target = lowd.GaussianMixture(num_mixes, z_dim, mu_target, logvar_target)
model = getattr(lowd, model_name)(target, z_dim, num_samples)
train(model)
def train_standardgaussian():
# mu_target = torch.FloatTensor([[-1,3],
# [0,3]])
# logvar_target = torch.FloatTensor([[-2,1],
# [-1,-2]])
# B = torch.eye(num_samples)
mu_target = torch.FloatTensor([0])
logvar_target = torch.FloatTensor([0])
target = lowd.GaussianMixture(num_mixes, z_dim, mu_target, logvar_target)
model = getattr(lowd, model_name)(target,
z_dim,
num_samples,
FreezeMean=False)
train(model)
def train_standardgaussian():
# mu_target = torch.FloatTensor([[-1,3],
# [0,3]])
# logvar_target = torch.FloatTensor([[-2,1],
# [-1,-2]])
# B = torch.eye(num_samples)
torch.manual_seed(123)
# mu_target = torch.cat([5 * torch.rand(10, 1) - 1, 5 * torch.rand(10, 1) + 5], 1)
# logvar_target = torch.cat([-2 * torch.rand(10, 1), torch.rand(10, 1)], 1)
# print(mu_target)
# print(logvar_target)
mu_target = torch.FloatTensor([0])
logvar_target = torch.FloatTensor([0])
target = lowd.GaussianMixture(num_mixes, z_dim, mu_target, logvar_target)
model = getattr(lowd, model_name)(target, z_dim, num_samples)
train(model)