本文整理汇总了Python中theano.tensor.matrices函数的典型用法代码示例。如果您正苦于以下问题:Python matrices函数的具体用法?Python matrices怎么用?Python matrices使用的例子?那么恭喜您, 这里精选的函数代码示例或许可以为您提供帮助。
在下文中一共展示了matrices函数的15个代码示例,这些例子默认根据受欢迎程度排序。您可以为喜欢或者感觉有用的代码点赞,您的评价将有助于系统推荐出更棒的Python代码示例。
示例1: __init__
def __init__(self,
gen_params, # dictionary of generative model parameters
GEN_MODEL, # class that inherits from GenerativeModel
rec_params, # dictionary of approximate posterior ("recognition model") parameters
REC_MODEL, # class that inherits from RecognitionModel
xDim=2, # dimensionality of latent state
yDim=2 # dimensionality of observations
):
# instantiate rng's
self.srng = RandomStreams(seed=234)
self.nrng = np.random.RandomState(124)
#---------------------------------------------------------
## actual model parameters
self.X, self.Y = T.matrices('X','Y') # symbolic variables for the data
self.xDim = xDim
self.yDim = yDim
# instantiate our prior & recognition models
self.mrec = REC_MODEL(rec_params, self.Y, self.xDim, self.yDim, self.srng, self.nrng)
self.mprior = GEN_MODEL(gen_params, self.xDim, self.yDim, srng=self.srng, nrng = self.nrng)
self.isTrainingRecognitionModel = True;
self.isTrainingGenerativeModel = True;
示例2: _gpu_matrix_dot
def _gpu_matrix_dot(matrix_a, matrix_b, matrix_c=None):
"""
Performs matrix multiplication.
Attempts to use the GPU if it's available. If the matrix multiplication
is too big to fit on the GPU, this falls back to the CPU after throwing
a warning.
Parameters
----------
matrix_a : WRITEME
matrix_b : WRITEME
matrix_c : WRITEME
"""
if not hasattr(ZCA._gpu_matrix_dot, 'theano_func'):
ma, mb = T.matrices('A', 'B')
mc = T.dot(ma, mb)
ZCA._gpu_matrix_dot.theano_func = \
theano.function([ma, mb], mc, allow_input_downcast=True)
theano_func = ZCA._gpu_matrix_dot.theano_func
try:
if matrix_c is None:
return theano_func(matrix_a, matrix_b)
else:
matrix_c[...] = theano_func(matrix_a, matrix_b)
return matrix_c
except MemoryError:
warnings.warn('Matrix multiplication too big to fit on GPU. '
'Re-doing with CPU. Consider using '
'THEANO_FLAGS="device=cpu" for your next '
'preprocessor run')
return np.dot(matrix_a, matrix_b, matrix_c)
示例3: sample_parallel
def sample_parallel():
print "並列"
x, y = T.matrices("a", "b")
diff = x - y
abs_diff = abs(diff)
diff_sq = diff**2
# 2つの行列を入力, 3つの行列のベクトルを出力
f = theano.function([x, y], [diff, abs_diff, diff_sq])
print f([[0,1],[2,3]], [[10,11],[12,13]])
print
示例4: test_grad
def test_grad(self, cls_ofg):
x, y, z = T.matrices('xyz')
e = x + y * z
op = cls_ofg([x, y, z], [e])
f = op(x, y, z)
f = f - T.grad(T.sum(f), y)
fn = function([x, y, z], f)
xv = np.ones((2, 2), dtype=config.floatX)
yv = np.ones((2, 2), dtype=config.floatX) * 3
zv = np.ones((2, 2), dtype=config.floatX) * 5
assert np.all(11.0 == fn(xv, yv, zv))
示例5: test_grad
def test_grad(self):
x, y, z = T.matrices('xyz')
e = x + y * z
op = OpFromGraph([x, y, z], [e], mode='FAST_RUN', grad_depth=2)
f = op(x, y, z)
f = f - T.grad(T.sum(f), y)
fn = function([x, y, z], f)
xv = numpy.ones((2, 2), dtype=config.floatX)
yv = numpy.ones((2, 2), dtype=config.floatX)*3
zv = numpy.ones((2, 2), dtype=config.floatX)*5
assert numpy.all(11.0 == fn(xv, yv, zv))
示例6: test_connection_pattern
def test_connection_pattern(self, cls_ofg):
# Basic case
x, y, z = T.matrices('xyz')
out1 = x * y
out2 = y * z
op1 = cls_ofg([x, y, z], [out1, out2])
results = op1.connection_pattern(None)
expect_result = [[True, False],
[True, True],
[False, True]]
assert results == expect_result
# Graph with ops that don't have a 'full' connection pattern
# and with ops that have multiple outputs
m, n, p, q = T.matrices('mnpq')
o1, o2 = op1(m, n, p)
out1, out2 = op1(o1, q, o2)
op2 = cls_ofg([m, n, p, q], [out1, out2])
results = op2.connection_pattern(None)
expect_result = [[True, False],
[True, True],
[False, True],
[True, True]]
assert results == expect_result
# Inner graph where some computation doesn't rely on explicit inputs
srng = RandomStreams(seed=234)
rv_u = srng.uniform((2, 2))
x, y = T.matrices('xy')
out1 = x + rv_u
out2 = y + 3
out3 = 3 + rv_u
op3 = cls_ofg([x, y], [out1, out2, out3])
results = op3.connection_pattern(None)
expect_result = [[True, False, False],
[False, True, False],
[True, False, True]]
assert results == expect_result
示例7: test_grad_grad
def test_grad_grad(self):
x, y, z = T.matrices('xyz')
e = x + y * z
op = OpFromGraph([x, y, z], [e])
f = op(x, y, z)
f = f - T.grad(T.sum(f), y)
f = f - T.grad(T.sum(f), y)
fn = function([x, y, z], f)
xv = numpy.ones((2, 2), dtype=config.floatX)
yv = numpy.ones((2, 2), dtype=config.floatX) * 3
zv = numpy.ones((2, 2), dtype=config.floatX) * 5
assert numpy.allclose(6.0, fn(xv, yv, zv))
示例8: _build
def _build(self):
if self._debug:
theano.config.compute_test_value = 'warn'
X,W = T.matrices('X','W')
if self._debug:
X.tag.test_value = np.random.rand(3,1)
W.tag.test_value = np.random.rand(5,3)
Z = T.dot(W,X)
A = self._activation(Z)
self._fpropagate = function([X, W],A)
self._layers = []
self._generate_initial_weights()
示例9: __init__
def __init__(self, shape):
self.in_size, self.out_size = shape
self.W = init_weights(shape)
self.b = init_bias(self.out_size)
self.gW = init_gradws(shape)
self.gb = init_bias(self.out_size)
D, X = T.matrices("D", "X")
def _active(X):
return T.nnet.sigmoid(T.dot(X, self.W) + self.b)
self.active = theano.function(inputs = [X], outputs = _active(X))
def _derive(D, X):
return D * ((1 - X) * X)
self.derive = theano.function(
inputs = [D, X],
outputs = _derive(D, X)
)
def _propagate(D):
return T.dot(D, self.W.T)
self.propagate = theano.function(inputs = [D], outputs = _propagate(D))
x, dy = T.rows("x","dy")
updates_grad = [(self.gW, self.gW + T.dot(x.T, dy)),
(self.gb, self.gb + dy)]
self.grad = theano.function(
inputs = [x, dy],
updates = updates_grad
)
updates_clear = [
(self.gW, self.gW * 0),
(self.gb, self.gb * 0)]
self.clear_grad = theano.function(
inputs = [],
updates = updates_clear
)
lr = T.scalar()
t = T.scalar()
updates_w = [
(self.W, self.W - self.gW * lr / t),
(self.b, self.b - self.gb * lr / t)]
self.update = theano.function(
inputs = [lr, t],
updates = updates_w
)
示例10: test_straightforward
def test_straightforward(self):
x, y, z = T.matrices('xyz')
e = x + y * z
op = OpFromGraph([x, y, z], [e], mode='FAST_RUN')
f = op(x, y, z) - op(y, z, x) # (1+3*5=array of 16) - (3+1*5=array of 8)
fn = function([x, y, z], f)
xv = numpy.ones((2, 2), dtype=config.floatX)
yv = numpy.ones((2, 2), dtype=config.floatX)*3
zv = numpy.ones((2, 2), dtype=config.floatX)*5
#print function, function.__module__
#print fn.maker.fgraph.toposort()
fn(xv, yv, zv)
assert numpy.all(8.0 == fn(xv, yv, zv))
assert numpy.all(8.0 == fn(xv, yv, zv))
示例11: test_size_changes
def test_size_changes(self):
x, y, z = T.matrices('xyz')
e = T.dot(x, y)
op = OpFromGraph([x, y], [e], mode='FAST_RUN')
f = op(x, op(y, z))
fn = function([x, y, z], f)
xv = numpy.ones((2, 3), dtype=config.floatX)
yv = numpy.ones((3, 4), dtype=config.floatX)*3
zv = numpy.ones((4, 5), dtype=config.floatX)*5
res = fn(xv, yv, zv)
assert res.shape == (2, 5)
assert numpy.all(180.0 == res)
res = fn(xv, yv, zv)
assert res.shape == (2, 5)
assert numpy.all(180.0 == res)
示例12: test_size_changes
def test_size_changes(self, cls_ofg):
x, y, z = T.matrices('xyz')
e = T.dot(x, y)
op = cls_ofg([x, y], [e])
f = op(x, op(y, z))
fn = function([x, y, z], f)
xv = np.ones((2, 3), dtype=config.floatX)
yv = np.ones((3, 4), dtype=config.floatX) * 3
zv = np.ones((4, 5), dtype=config.floatX) * 5
res = fn(xv, yv, zv)
assert res.shape == (2, 5)
assert np.all(180.0 == res)
res = fn(xv, yv, zv)
assert res.shape == (2, 5)
assert np.all(180.0 == res)
示例13: test_straightforward
def test_straightforward(self, cls_ofg):
x, y, z = T.matrices('xyz')
e = x + y * z
op = cls_ofg([x, y, z], [e])
# (1+3*5=array of 16) - (3+1*5=array of 8)
f = op(x, y, z) - op(y, z, x)
fn = function([x, y, z], f)
xv = np.ones((2, 2), dtype=config.floatX)
yv = np.ones((2, 2), dtype=config.floatX) * 3
zv = np.ones((2, 2), dtype=config.floatX) * 5
# print function, function.__module__
# print fn.maker.fgraph.toposort()
fn(xv, yv, zv)
assert np.all(8.0 == fn(xv, yv, zv))
assert np.all(8.0 == fn(xv, yv, zv))
示例14: test_shared
def test_shared(self, cls_ofg):
x, y, z = T.matrices('xyz')
s = shared(np.random.rand(2, 2).astype(config.floatX))
e = x + y * z + s
op = cls_ofg([x, y, z], [e])
# (1+3*5=array of 16) - (3+1*5=array of 8)
f = op(x, y, z) - op(y, z, x)
fn = function([x, y, z], f)
xv = np.ones((2, 2), dtype=config.floatX)
yv = np.ones((2, 2), dtype=config.floatX) * 3
zv = np.ones((2, 2), dtype=config.floatX) * 5
# print function, function.__module__
# print fn.maker.fgraph.toposort()
assert np.allclose(8.0, fn(xv, yv, zv))
assert np.allclose(8.0, fn(xv, yv, zv))
示例15: __init__
def __init__(self, shape, X):
prefix = "Softmax_"
self.in_size, self.out_size = shape
self.W = init_weights(shape, prefix + "W")
self.b = init_bias(self.out_size, prefix + "b")
self.gW = init_gradws(shape, prefix + "gW")
self.gb = init_bias(self.out_size, prefix + "gb")
D = T.matrices("D")
self.X = X
def _active(X):
return T.nnet.softmax(T.dot(X, self.W) + self.b)
self.active = theano.function(inputs = [self.X], outputs = _active(self.X))
def _propagate(D):
return T.dot(D, self.W.T)
self.propagate = theano.function(inputs = [D], outputs = _propagate(D))
x, dy = T.rows("x","dy")
updates_grad = [(self.gW, self.gW + T.dot(x.T, dy)),
(self.gb, self.gb + dy)]
self.grad = theano.function(
inputs = [x, dy],
updates = updates_grad
)
updates_clear = [
(self.gW, self.gW * 0),
(self.gb, self.gb * 0)]
self.clear_grad = theano.function(
inputs = [],
updates = updates_clear
)
lr = T.scalar()
t = T.scalar()
updates_w = [
(self.W, self.W - self.gW * lr / t),
(self.b, self.b - self.gb * lr / t)]
self.update = theano.function(
inputs = [lr, t],
updates = updates_w
)
self.params = [self.W, self.b]