def constant(x, name=None, shape=None, dtype=None):
"""Initialize a tensor with constant value.
If dtype is ``None``, use ``config.floatX``.
Parameters
----------
x : basic numerical type
The constant value.
name : str or None
The name of Tensor.
shape : list, tuple or None
The shape of Tensor.
dtype : str or None
The data type of Tensor.
Returns
-------
Tensor
The initialized tensor.
"""
if dtype is None: dtype = config.floatX
else:
if dtype not in _DATA_TYPES.keys():
raise TypeError("Unsupported data type: {}".format(dtype))
if shape is None: shape = ()
np_value = x * np.ones(shape, dtype=_DATA_TYPES[dtype])
output = Tensor(name=name, shape=shape, dtype=dtype)
output.set_value(np_value)
return output
def zeros(shape, dtype=None):
"""Initialize a tensor with zeros.
If dtype is ``None``, use ``config.floatX``.
Parameters
----------
shape : tuple or list
The shape of Tensor.
dtype : str or None
The data type of Tensor.
Returns
-------
Tensor
The initialized tensor.
"""
if dtype is None: dtype = config.floatX
else:
if dtype not in _DATA_TYPES.keys():
raise TypeError("Unsupported data type: {}".format(dtype))
np_value = np.zeros(shape, dtype=_DATA_TYPES[dtype])
output = Tensor(shape=shape, dtype=dtype)
output.set_value(np_value)
return output
def ones(shape, dtype=None):
"""Initialize a tensor with ones.
If dtype is ``None``, use ``config.floatX``.
Parameters
----------
shape : tuple or list
The shape of Tensor.
dtype : str or None
The data type of Tensor.
Returns
-------
Tensor
The initialized tensor.
"""
if dtype is None: dtype = config.floatX
else:
if dtype not in _DATA_TYPES.keys():
raise TypeError("Unsupported data type: {}".format(dtype))
np_value = np.ones(shape, dtype=_DATA_TYPES[dtype])
output = Tensor(shape=shape, dtype=dtype)
output.set_value(np_value)
return output
def constant(x, name=None, shape=None, dtype=None):
"""Initialize a tensor with constant value.
If dtype is ``None``, use ``config.floatX``.
Parameters
----------
x : basic numerical type
The constant value.
name : str or None
The name of Tensor.
shape : list, tuple or None
The shape of Tensor.
dtype : str or None
The data type of Tensor.
Returns
-------
Tensor
The initialized tensor.
"""
if dtype is None: dtype = config.floatX
else:
if dtype not in _DATA_TYPES.keys():
raise TypeError("Unsupported data type: {}".format(dtype))
if shape is None: shape = ()
np_value = x * np.ones(shape, dtype=_DATA_TYPES[dtype])
output = Tensor(name=name, shape=shape, dtype=dtype)
output.set_value(np_value)
return output
def convert_to_tensor(value, dtype=None, name=None, **kwargs):
"""Converts the given value to a Tensor.
Parameters
----------
value : basic type, list or numpy.ndarray
The value to convert.
dtype : Dtype or None
The data type. If ``None``, inferred from the type of `value`.
name : str or None
The Optional name.
Returns
-------
Tensor
The output tensor.
"""
if dtype is not None:
if not isinstance(dtype, str):
if isinstance(dtype, dtypes.DType):
dtype = dtype.name
else:
raise ValueError('The dtype should be a str of a tf.Dtype.')
tensor = Tensor(name=name, dtype=dtype)
tensor.set_value(value)
return tensor
def convert_to_tensor(value, dtype=None, name=None, **kwargs):
"""Converts the given value to a Tensor.
Parameters
----------
value : basic type, list or numpy.ndarray
The value to convert.
dtype : Dtype or None
The data type. If ``None``, inferred from the type of `value`.
name : str or None
The Optional name.
Returns
-------
Tensor
The output tensor.
"""
if dtype is not None:
if not isinstance(dtype, str):
if isinstance(dtype, dtypes.DType):
dtype = dtype.name
else:
raise ValueError('The dtype should be a str of a tf.Dtype.')
tensor = Tensor(name=name, dtype=dtype)
tensor.set_value(value)
return tensor
def WrapConstants(constants, dtype='float32'):
if not isinstance(constants, Tensor):
if not isinstance(constants, np.ndarray):
constants = np.array(constants, dtype=dtype)
tensor = Tensor()
tensor.set_value(constants)
tensor.shape = constants.shape
constants = tensor
return constants
def _set_param(self, layer_id, param_id, param_type, param):
if not isinstance(param, Tensor):
if isinstance(param, np.ndarray):
paramT = Tensor('/tmp/rnn_param').Variable()
paramT.set_value(param)
param = paramT
else:
raise ValueError('Excepted a tensor or numpy array.')
self.weights.expressions = dict() # Clear cached expressions
outputs = RNNParamSet([self.weights, param],
layer_id,
param_id,
param_type,
rnn_mode=self.mode,
input_size=self.input_size,
hidden_size=self.hidden_size,
num_layers=self.num_layers,
num_directions=self.num_directions)
for k, v in outputs.expressions.items():
dg.workspace.RunOperator(v)
