Theano Tutorial - 03. TensorVariable

From Deep_Learning_Machine_Learning_and_Artificial_Intelligence
Revision as of 01:38, 27 May 2016 by Kai (Talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

There are three types of variable: TensorVariable, TensorConstant, and TensorSharedVariable.

class theano.tensor.TensorVariable(Variable, _tensor_py_operators)
class theano.tensor.TensorConstant(Variable, _tensor_py_operators)
class theano.tensor.TensorSharedVariable(Variable, _tensor_py_operators)

When we create a tensor, it is an instance of TensorVariable class, with type field has information to TensorType.

>>> a = T.scalar()
>>> print(a)
<TensorType(float64, scalar)>
>>> print(a.type)
TensorType(float64, scalar)
>>> print(type(a))
<class 'theano.tensor.var.TensorVariable'>

Tensor constant is created with theano.tensor.constant()

>>> const = T.constant(100)
>>> print(const)
TensorConstant{100}
>>> print(const.type)
TensorType(int8, scalar)
>>> print(type(const))
<class 'theano.tensor.var.TensorConstant'>

Shared variable is created out of numpy ndarray with theano.shared()

>>> b = theano.shared(np.asarray([[1,2],[3,4]]))
>>> print(b)
<TensorType(int64, matrix)>
>>> print(b.type)
TensorType(int64, matrix)
>>> print(type(b))
<class 'theano.tensor.sharedvar.TensorSharedVariable'>
Personal tools
Namespaces

Variants
Actions
Navigation
Toolbox