def set_hidden_state(self, state):
if len(state) != len(self.state):
raise Exception("Provided hidden state array does not match layer hidden states")
for s, h in zip(self.state, state):
if s.eval().shape != h.shape:
raise Exception("Hidden state shape not compatible")
s.set_value(floatX(h))
def set_hidden_state(self, state):
if len(state) != len(self.state):
raise Exception(
"Provided hidden state array does not match layer hidden states"
)
for s, h in zip(self.state, state):
if s.eval().shape != h.shape:
raise Exception("Hidden state shape not compatible")
s.set_value(floatX(h))
def get_hsn_features(self, split, data_key):
""" Return image features vector for split[data_key]."""
try:
return floatX(self.source_dataset[split][data_key]
['final_hidden_features'])
except KeyError:
# this image -- description pair doesn't have a source-language
# vector. Raise a KeyError so the requester can deal with the
# missing data.
logger.warning("Skipping '%s' because it doesn't have a source vector", data_key)
raise KeyError
def set_weights(self, weights):
np = len(self.params)
nw = len(weights)
ns = len(self.state)
if nw == np + ns:
nw = np
state = weights[-ns:]
self.set_hidden_state(state)
params = self.params[:np]
weights = weights[:nw]
assert len(params) == len(weights), 'Provided weight array does not match layer weights (' + \
str(len(params)) + ' layer params vs. ' + str(len(weights)) + ' provided weights)'
for p, w in zip(params, weights):
if p.eval().shape != w.shape:
raise Exception("Layer shape %s not compatible with weight shape %s." % (p.eval().shape, w.shape))
p.set_value(floatX(w))
def set_weights(self, weights):
np = len(self.params)
nw = len(weights)
ns = len(self.state)
if nw == np + ns:
nw = np
state = weights[-ns:]
self.set_hidden_state(state)
params = self.params[:np]
weights = weights[:nw]
assert len(params) == len(weights), 'Provided weight array does not match layer weights (' + \
str(len(params)) + ' layer params vs. ' + str(len(weights)) + ' provided weights)'
for p, w in zip(params, weights):
if p.eval().shape != w.shape:
raise Exception(
"Layer shape %s not compatible with weight shape %s." %
(p.eval().shape, w.shape))
p.set_value(floatX(w))
def set_weights(self, weights):
for p, w in zip(self.params, weights):
p.set_value(floatX(w))
def set_weights(self, weights):
self.running_mean.set_value(floatX(weights[-2]))
self.running_std.set_value(floatX(weights[-1]))
super(BatchNormalization, self).set_weights(weights[:-2])
def set_state(self, value_list):
assert len(self.updates) == len(value_list)
for u, v in zip(self.updates, value_list):
u[0].set_value(floatX(v))
def set_state(self, value_list):
assert len(self.updates) == len(value_list)
for u, v in zip(self.updates, value_list):
u[0].set_value(floatX(v))