def transform(self, loader):
'''Apply the model on the provided data loader.
Arguments:
loader (DataLoader): the data you wish to transform
'''
latent = []
for x, _ in loader:
x = self.transformer.x(x)
latent.append(x.mm(self.encoder_matrix.t()))
return _cat(latent)
def transform(self, loader):
'''Apply the model on the provided data loader.
Arguments:
loader (DataLoader): the data you wish to transform
'''
self.eval()
latent = []
for x, _ in loader:
x = self.transformer.x(x)
if self.use_cuda:
x = x.cuda(non_blocking=self.non_blocking)
y = self.encode(x)
if self.cuda:
y = y.cpu()
latent.append(y)
return _cat(latent).data
def transform(self, loader):
'''Apply the model on the provided data loader.
Arguments:
loader (DataLoader): the data you wish to transform
'''
self.eval()
latent = []
for i, (x, _) in enumerate(loader):
x = self.transformer.x(
x, variable=True, volatile=True, requires_grad=False)
if self.use_cuda:
x = x.cuda(async=self.async)
y = self.encode(x)
if self.cuda:
y = y.cpu()
latent.append(y)
return _cat(latent).data
def transform(self, loader):
'''Apply the model on the provided data loader.
Arguments:
loader (DataLoader): the data you wish to transform
'''
self.eval()
latent = []
for i, (x, _) in enumerate(loader):
x = self.transformer.x(x,
variable=True,
volatile=True,
requires_grad=False)
if self.use_cuda:
x = x.cuda()
try:
x = self.batch_normalization(x)
except AttributeError:
pass
y = self.encode(x)
if self.cuda:
y = y.cpu()
latent.append(y)
return _cat(latent).data
def __call__(self, **kwargs):
selection = kwargs.setdefault('selection', None)
feature_data, ref_data = self.bootstrap(selection=selection)
ref_data = _cat(ref_data)
dsc_data = self.discretizer(ref_data)
return feature_data, ref_data, dsc_data