Skip to content

P
pytensor
提交 5548fee1 authored 作者: Razvan Pascanu's avatar Razvan Pascanu
浏览文件
操作

removed tests for things that are going to get depricated; added name to

variables and some other cosmetic changes.
上级 97bd98fd
隐藏空白字符变更
内嵌 并排
正在显示 包含 19 行增加 和 78 行删除
theano/scan_module/tests/test_scan.py
浏览文件 @ 5548fee1
......@@ -383,9 +383,6 @@ class T_Scan(unittest.TestCase):
assert len(scan_node) == 1
scan_node = scan_node[0]
scan_node_topo = scan_node.op.fn.maker.env.toposort()
#theano.printing.pydotprint(f2, outfile='out2.png', high_contrast=True)
#theano.printing.pydotprint(scan_node.op.fn,
# outfile='inner2.png', high_contrast=True)
# check that there is no gpu transfer in the inner loop.
assert any([isinstance(node.op, theano.sandbox.cuda.GpuElemwise) for node in scan_node_topo])
......@@ -1437,8 +1434,8 @@ class T_Scan(unittest.TestCase):
assert numpy.allclose(f([2,3]) , 5)
def test_computing_gradient(self):
x1 = theano.tensor.scalar()
x2 = theano.shared(numpy.array([1,2,3,4,5]))
x1 = theano.tensor.scalar('x1')
x2 = theano.shared(numpy.array([1,2,3,4,5]), name = 'x2')
K = x2*x1
out,updates = theano.scan(lambda i,v: theano.tensor.grad(K[i], v),
......@@ -1494,6 +1491,9 @@ class T_Scan(unittest.TestCase):
iteration.
This unit test addresses the bug fix of changeset ba7157e95cb1.
!!! test lost some of its meaning because return_steps has been
depricated !!!
"""
a = theano.tensor.vector()
init_a = theano.tensor.vector()
......@@ -1503,8 +1503,9 @@ class T_Scan(unittest.TestCase):
return a+1, {b:2*b}
out, updates = theano.scan(inner_func,
outputs_info = [{'initial': init_a, 'return_steps': 1}],
outputs_info = [{'initial': init_a}],
n_steps=1)
out = out[-1]
assert out.type.ndim == a.type.ndim
assert updates[b].type.ndim == b.type.ndim
......@@ -1717,69 +1718,6 @@ class T_Scan(unittest.TestCase):
assert numpy.allclose(theano_y , v_y)
### TEST store steps / return steps
def test_return_steps(self):
rng = numpy.random.RandomState(utt.fetch_seed())
vW_in2 = asarrayX(rng.uniform(size = (2,), low = -5.,high = 5.))
vW = asarrayX(rng.uniform(size = (2,2), low = -5.,high = 5.))
vWout = asarrayX(rng.uniform(size = (2,), low = -5.,high = 5.))
vW_in1 = asarrayX(rng.uniform(size = (2,2), low = -5.,high = 5.))
v_u1 = asarrayX(rng.uniform(size = (8,2), low = -5., high = 5.))
v_u2 = asarrayX(rng.uniform(size = (8,), low = -5.,high = 5.))
v_x0 = asarrayX(rng.uniform(size = (2,), low = -5.,high = 5.))
v_y0 = asarrayX(rng.uniform(size = (3,)))
W_in2 = theano.shared(vW_in2, name='win2')
W = theano.shared(vW, name='w')
W_out = theano.shared(vWout, name = 'wout')
W_in1 = theano.tensor.matrix('win')
u1 = theano.tensor.matrix('u1')
u2 = theano.tensor.vector('u2')
x0 = theano.tensor.vector('x0')
y0 = theano.tensor.vector('y0')
def f_rnn_cmpl(u1_t, u2_t, x_tm1, y_tm1, y_tm3, W_in1):
return [y_tm3+1, theano.dot(u1_t,W_in1) + u2_t * W_in2 + \
theano.dot(x_tm1, W),
y_tm1 + theano.dot(x_tm1, W_out)]
outputs, updates = theano.scan( f_rnn_cmpl
, [ u1
, u2]
, [ dict(store_steps = 3)
, dict(initial = x0, return_steps = 2)
, dict(initial=y0, taps=[-1,-3],
return_steps = 4)]
, W_in1
, n_steps = None
, truncate_gradient = -1
, go_backwards = False)
f4 = theano.function([u1,u2,x0,y0,W_in1], outputs
, updates = updates
, allow_input_downcast = True
)
# compute the values in numpy
v_x = numpy.zeros((8,2),dtype=theano.config.floatX)
v_y = numpy.zeros((8,),dtype=theano.config.floatX)
v_x[0] = numpy.dot(v_u1[0],vW_in1) + v_u2[0]*vW_in2 + \
numpy.dot(v_x0,vW)
v_y[0] = numpy.dot(v_x0,vWout) + v_y0[2]
for i in xrange(1,8):
v_x[i] = numpy.dot(v_u1[i],vW_in1) + v_u2[i]*vW_in2 + \
numpy.dot(v_x[i-1],vW)
v_y[i] = numpy.dot(v_x[i-1], vWout) + v_y[i-1]
(theano_dump, theano_x,theano_y) = f4( v_u1, v_u2, v_x0, v_y0, vW_in1)
assert numpy.allclose(theano_x , v_x[-2:])
assert numpy.allclose(theano_y , v_y[-4:])
def test_scan_as_tensor_on_gradients(self):
......@@ -1832,19 +1770,20 @@ class T_Scan(unittest.TestCase):
theano.dot(x_tm1, W),
y_tm1 + theano.dot(x_tm1, W_out)]
outputs, updates = theano.scan( f_rnn_cmpl
_outputs, updates = theano.scan( f_rnn_cmpl
, [ u1
, u2]
, [ dict(return_steps = 1)
, dict(initial = x0
, return_steps = 1)
, dict(initial=y0, taps=[-1,-3],
return_steps = 1)]
, [ None
, dict(initial = x0)
, dict(initial=y0, taps=[-1,-3])]
, W_in1
, n_steps = None
, truncate_gradient = -1
, go_backwards = False)
outputs = [ \
_outputs[0][-1],
_outputs[1][-1],
_outputs[2][-1] ]
f4 = theano.function([u1,u2,x0,y0,W_in1], outputs
, updates = updates
, allow_input_downcast = True
......@@ -1989,7 +1928,7 @@ class T_Scan(unittest.TestCase):
# The following test will fail in DebugMode if there are
# some problems in Scan.infer_shape
def test_remove_stuff(self):
x = theano.tensor.vector()
x = theano.tensor.vector('x')
def lm(m):
trng = theano.tensor.shared_randomstreams.RandomStreams(
......@@ -2002,6 +1941,7 @@ class T_Scan(unittest.TestCase):
sequences = x,
n_steps = None,
truncate_gradient = -1,
name ='forward',
go_backwards = False)
go1 = theano.tensor.grad(o1.mean(), wrt = x)
f = theano.function([x],go1, updates = updates,
......@@ -2010,9 +1950,10 @@ class T_Scan(unittest.TestCase):
#theano.printing.debugprint(f, print_type=True)
topo = f.maker.env.toposort()
# this new assert is here to test if scan_merging works ..
nb_scan = len([n for n in topo
if isinstance(n.op, theano.scan_module.scan_op.Scan)])
self.assertTrue(nb_scan == 2)
self.assertTrue(nb_scan == 1)
nb_shape_i = len([n for n in topo
if isinstance(n.op, theano.tensor.opt.Shape_i)])
if theano.config.mode != 'FAST_COMPILE':
......
Markdown 格式
0% 或
您添加了 0 人 到此讨论。请谨慎行事。
请先完成此评论的编辑！
请 注册 或者 后发表评论