def check_tensor_shape(tensor1, tensor2, name1=None, name2=None):
if name1 is None: name1 = misc.retrieve_name(tensor1)
if name2 is None: name2 = misc.retrieve_name(tensor2)
assert isinstance(name1, str) and isinstance(name2, str)
assert isinstance(tensor1, tf.Tensor) and isinstance(tensor2, tf.Tensor)
shape1, shape2 = tensor1.shape.as_list(), tensor2.shape.as_list()
if shape1 != shape2:
raise ValueError(
'!! {}.shape({}) should be equal with {}.shape({}) '.format(
name1, shape1, name2, shape2))
def check_callable(f, name=None, allow_none=True):
if name is None: name = misc.retrieve_name(f)
flag = True
if not allow_none and f is None: flag = False
if f is not None and not callable(f): flag = False
if flag: return f
else: raise TypeError('!! {} must be callable'.format(name))
def eval_show(tensor, name=None, feed_dict=None):
if name is None: name = misc.retrieve_name(tensor)
sess = tf.get_default_session()
val = sess.run(tensor, feed_dict=feed_dict)
if len(val.shape) > 1:
show_status('{} = '.format(name))
pprint(val)
else:
show_status('{} = {}'.format(name, val))
return val
def check_positive_integer(x, allow_zero=False, name=None):
if not isinstance(x, int) or x < 0 or not allow_zero and x == 0:
if name is None: name = misc.retrieve_name(x)
raise ValueError('!! {} must be a positive integer'.format(name))
return x