def generate(params):
dbn_params = dbn.stack_params(params)
# 10 fantasies.
# initial_pixels = np.zeros((10, 28**2))
# initial_pixels = inputs[0:10]
# Clamp the softmax units, one for each class.
sample_v_softmax_clamped = functools.partial(sample_v_softmax,
labels=np.eye(10))
# Perform an upward pass from the pixels to the visible units of
# the top-level RBM.
# initial_v = np.hstack((
# np.eye(10),
# up_pass(dbn_params, initial_pixels)))
initial_v = np.hstack(
(np.eye(10), np.random.random((10, dbn_params[-1].W.shape[1] - 10))))
# Initialize the gibbs chain.
gc = rbm.gibbs_chain(initial_v, dbn_params[-1], rbm.sample_h,
sample_v_softmax_clamped)
tile_2_by_5 = functools.partial(utils.tile, grid_shape=(2, 5))
gen = itertools.islice(gc, 1, None, 1)
gen = itertools.islice(gen, 2000)
gen = itertools.imap(operator.itemgetter(1), gen)
gen = itertools.imap(lambda v: down_pass(dbn_params, v), gen)
gen = itertools.imap(tile_2_by_5, gen)
# Save to disk.
utils.save_images(gen, tempfile.mkdtemp(dir=OUTPUT_PATH))
def generate(params, n=1, start=1, step=20, count=100):
# Use the probability as pixel intensities.
# This means we can't use the very first sample from the chain, as
# it has v_mean == None.
assert start > 0
num_vis = params.W.shape[1]
#initial_v = np.zeros((n, num_vis))
initial_v = inputs[:n]
#initial_v = np.random.random((n, num_vis))
gc = rbm.gibbs_chain(initial_v, params, rbm.sample_h, rbm.sample_v)
g = itertools.islice(gc, start, None, step)
g = itertools.islice(g, count)
g = itertools.imap(operator.itemgetter(1), g)
g = itertools.imap(utils.tile, g)
utils.save_images(g, tempfile.mkdtemp(dir=OUTPUT_PATH))
def generate(params, n=1, start=1, step=20, count=100):
# Use the probability as pixel intensities.
# This means we can't use the very first sample from the chain, as
# it has v_mean == None.
assert start > 0
num_vis = params.W.shape[1]
#initial_v = np.zeros((n, num_vis))
initial_v = inputs[:n]
#initial_v = np.random.random((n, num_vis))
gc = rbm.gibbs_chain(initial_v,
params,
rbm.sample_h,
rbm.sample_v)
g = itertools.islice(gc, start, None, step)
g = itertools.islice(g, count)
g = itertools.imap(operator.itemgetter(1), g)
g = itertools.imap(utils.tile, g)
utils.save_images(g, tempfile.mkdtemp(dir=OUTPUT_PATH))
def generate(params):
dbn_params = dbn.stack_params(params)
# 10 fantasies.
# initial_pixels = np.zeros((10, 28**2))
# initial_pixels = inputs[0:10]
# Clamp the softmax units, one for each class.
sample_v_softmax_clamped = functools.partial(sample_v_softmax,
labels=np.eye(10))
# Perform an upward pass from the pixels to the visible units of
# the top-level RBM.
# initial_v = np.hstack((
# np.eye(10),
# up_pass(dbn_params, initial_pixels)))
initial_v = np.hstack((
np.eye(10),
np.random.random((10, dbn_params[-1].W.shape[1] - 10))))
# Initialize the gibbs chain.
gc = rbm.gibbs_chain(initial_v,
dbn_params[-1],
rbm.sample_h,
sample_v_softmax_clamped)
tile_2_by_5 = functools.partial(utils.tile, grid_shape=(2, 5))
gen = itertools.islice(gc, 1, None, 1)
gen = itertools.islice(gen, 2000)
gen = itertools.imap(operator.itemgetter(1), gen)
gen = itertools.imap(lambda v: down_pass(dbn_params, v), gen)
gen = itertools.imap(tile_2_by_5, gen)
# Save to disk.
utils.save_images(gen, tempfile.mkdtemp(dir=OUTPUT_PATH))
