Skip to content

P
pytensor
提交 ce690a69 authored 作者: Razvan Pascanu's avatar Razvan Pascanu 提交者： David Warde-Farley
浏览文件
操作

Made the code pep8 compatible

上级 8c08c76e
隐藏空白字符变更
内嵌 并排
正在显示 包含 100 行增加 和 118 行删除
theano/ifelse.py
浏览文件 @ ce690a69
"""
IfElse is an Op that works with the LazyLinker to support conditional graph evaluation.
IfElse is an Op that works with the LazyLinker to support conditional graph
evaluation.
:TODO: Add text to library documentation describing the IfElse Op.
"""
__docformat__ = 'restructedtext en'
__authors__ = ( "Razvan Pascanu "
"James Bergstra "
"Dumitru Erhan ")
__authors__ = ("Razvan Pascanu "
"James Bergstra "
"Dumitru Erhan ")
__copyright__ = "(c) 2010, Universite de Montreal"
__contact__ = "Razvan Pascanu <r.pascanu@gmail>"
......@@ -27,8 +28,6 @@ from scan_module.scan_utils import clone
_logger = logging.getLogger('theano.ifelse')
class IfElse(PureOp):
"""
Op that works with CVM/VM to support conditional graph evaluation. Note,
......@@ -47,20 +46,19 @@ class IfElse(PureOp):
False branch before picking one.
"""
def __init__(self, n_outs, as_view=False, gpu = False, name = None):
def __init__(self, n_outs, as_view=False, gpu=False, name=None):
if as_view:
# check destroyhandler and others to ensure that a view_map with
# multiple inputs can work
view_map = {}
for idx in xrange(n_outs):
view_map[idx] = [idx+1]
view_map[idx] = [idx + 1]
self.view_map = view_map
#raise NotImplementedError('Cond must copy for now')
self.as_view = as_view
self.gpu = gpu
self.gpu = gpu
self.n_outs = n_outs
self.name = name
self.name = name
def __eq__(self, other):
if not type(self) == type(other):
......@@ -75,7 +73,7 @@ class IfElse(PureOp):
def __hash__(self):
rval = ( hash(type(self)) ^
rval = (hash(type(self)) ^
hash(self.as_view) ^
hash(self.gpu) ^
hash(self.n_outs))
......@@ -83,7 +81,7 @@ class IfElse(PureOp):
return rval
def __str__(self):
name ='if{%s'%str(self.name)
name = 'if{%s' % str(self.name)
if self.as_view:
name += ',inplace'
if self.gpu:
......@@ -91,7 +89,6 @@ class IfElse(PureOp):
name += '}'
return name
def infer_shape(self, node, inputs_shapes):
# By construction, corresponding then/else pairs have the same number
# of dimensions
......@@ -103,7 +100,7 @@ class IfElse(PureOp):
new_ts_inputs += list(ts_shape)
else:
# It can be None for generic objects
return [None]*self.n_outs
return [None] * self.n_outs
new_fs_inputs = []
for fs_shape in fs_shapes:
......@@ -111,18 +108,18 @@ class IfElse(PureOp):
new_fs_inputs += list(fs_shape)
else:
# It can be None for generic objects
return [None]*self.n_outs
return [None] * self.n_outs
assert len(new_ts_inputs) == len(new_fs_inputs)
if len(new_ts_inputs + new_fs_inputs) > 0:
new_ifelse = IfElse(
n_outs = len(new_ts_inputs),
as_view = False,
gpu = False,
name='shape_'+str(self.name))
n_outs=len(new_ts_inputs),
as_view=False,
gpu=False,
name='shape_' + str(self.name))
new_outs = new_ifelse.make_node(node.inputs[0],
*(new_ts_inputs+new_fs_inputs)).outputs
*(new_ts_inputs + new_fs_inputs)).outputs
else:
new_outs = []
......@@ -138,12 +135,8 @@ class IfElse(PureOp):
out_shapes += [tuple(current_shape)]
return out_shapes
def R_op(self, inputs, eval_points):
return self.make_node(inputs[0],*eval_points[1:]).outputs
return self.make_node(inputs[0], *eval_points[1:]).outputs
def make_node(self, c, *args):
if not self.gpu:
......@@ -160,15 +153,15 @@ class IfElse(PureOp):
ts = args[:self.n_outs]
fs = args[self.n_outs:]
for t,f in zip(ts, fs):
for t, f in zip(ts, fs):
if t.type != f.type:
raise TypeError(('IfElse requires same types for true and '
'false return values'), t, f, t.type, f.type)
if c.ndim >0:
if c.ndim > 0:
raise TypeError(('Condition given to the op has to be a scalar '
'with 0 standing for False, anything else for True'))
return Apply(self, [c]+list(args), [t.type() for t in ts])
'with 0 standing for False, anything else '
'for True'))
return Apply(self, [c] + list(args), [t.type() for t in ts])
def grad(self, ins, grads):
ts = ins[1:][:self.n_outs]
......@@ -179,35 +172,31 @@ class IfElse(PureOp):
else:
nw_name_t = None
nw_name_f = None
if_true_op = IfElse(n_outs = self.n_outs,
as_view = self.as_view,
gpu = self.gpu,
name = nw_name_t)
if_true_op = IfElse(n_outs=self.n_outs,
as_view=self.as_view,
gpu=self.gpu,
name=nw_name_t)
if_false_op = IfElse(n_outs = self.n_outs,
as_view = self.as_view,
gpu = self.gpu,
name = nw_name_f)
if_false_op = IfElse(n_outs=self.n_outs,
as_view=self.as_view,
gpu=self.gpu,
name=nw_name_f)
if_true = ([ins[0]]+ grads+ [theano.tensor.zeros_like(t)
if_true = ([ins[0]] + grads + [theano.tensor.zeros_like(t)
for t in ts])
if_false = ([ins[0]] + [theano.tensor.zeros_like(f)
for f in fs] + grads)
return ([None]+
return ([None] +
if_true_op.make_node(*if_true).outputs +
if_false_op.make_node(*if_false).outputs )
if_false_op.make_node(*if_false).outputs)
def make_thunk(self, node, storage_map, compute_map, no_recycling):
outtypes = [ out.type for out in node.outputs]
cond = node.inputs[0]
outtypes = [out.type for out in node.outputs]
cond = node.inputs[0]
ts = node.inputs[1:][:self.n_outs]
fs = node.inputs[1:][self.n_outs:]
outputs = node.outputs
outputs = node.outputs
def thunk():
if not compute_map[cond][0]:
......@@ -215,7 +204,7 @@ class IfElse(PureOp):
else:
truthval = storage_map[cond][0]
if truthval != 0:
ls = [idx+1 for idx in xrange(self.n_outs)
ls = [idx + 1 for idx in xrange(self.n_outs)
if not compute_map[ts[idx]][0]]
if len(ls) > 0:
return ls
......@@ -232,7 +221,7 @@ class IfElse(PureOp):
storage_map[out][0] = oval
return []
else:
ls = [1+idx+self.n_outs for idx in xrange(self.n_outs)
ls = [1 + idx + self.n_outs for idx in xrange(self.n_outs)
if not compute_map[fs[idx]][0]]
if len(ls) > 0:
return ls
......@@ -249,12 +238,12 @@ class IfElse(PureOp):
return []
thunk.lazy = True
thunk.inputs = [ storage_map[v] for v in node.inputs]
thunk.outputs = [ storage_map[v] for v in node.outputs]
thunk.inputs = [storage_map[v] for v in node.inputs]
thunk.outputs = [storage_map[v] for v in node.outputs]
return thunk
def ifelse( cond, true_branch, false_branch, name = None):
def ifelse(cond, true_branch, false_branch, name=None):
"""
This function corresponds to a if statement, returning inputs in the
``true_branch`` if ``cond`` evaluates to True or inputs in the
......@@ -293,34 +282,33 @@ def ifelse( cond, true_branch, false_branch, name = None):
false_branch = [false_branch]
if len(true_branch) != len(false_branch):
raise ValueError(( 'The number of values on the `then` branch'+
' should have the same number of variables as '+
'the `else` branch : (variables on `then` '+
'%d'%len(true_branch)+ ', variables on `else` '+
'%d'%len(false_branch)+')'))
raise ValueError(('The number of values on the `then` branch'
' should have the same number of variables as '
'the `else` branch : (variables on `then` '
'%d' % len(true_branch) + ', variables on `else` '
'%d' % len(false_branch) + ')'))
new_ifelse = IfElse(n_outs = len(true_branch),
new_ifelse = IfElse(n_outs=len(true_branch),
as_view=False,
gpu = False,
name = name)
gpu=False,
name=name)
ins = [cond] + list(true_branch) + list(false_branch)
rval = new_ifelse.make_node(*ins).outputs
if type(rval) in (list,tuple) and len(rval) == 1:
if type(rval) in (list, tuple) and len(rval) == 1:
return rval[0]
else:
return rval
@gof.local_optimizer([None])
def cond_make_inplace(node):
op = node.op
if isinstance(op, IfElse) and not op.as_view :
return IfElse(n_outs = op.n_outs,
as_view = True,
gpu = op.gpu,
name = op.name ).make_node(*node.inputs).outputs
if isinstance(op, IfElse) and not op.as_view:
return IfElse(n_outs=op.n_outs,
as_view=True,
gpu=op.gpu,
name=op.name).make_node(*node.inputs).outputs
return False
optdb.register('cond_make_inplace', opt.in2out(cond_make_inplace,
......@@ -330,29 +318,30 @@ ifelse_equilibrium = gof.EquilibriumDB()
ifelse_seqopt = gof.SequenceDB()
ifelse_equilibrium.register('seq_ifelse', ifelse_seqopt, 'fast_run',
'ifelse')
optdb.register('ifelse_equilibriumOpt', ifelse_equilibrium, .5, 'fast_run', 'ifelse')
optdb.register('ifelse_equilibriumOpt', ifelse_equilibrium, .5, 'fast_run',
'ifelse')
@gof.local_optimizer([None])
def cond_merge_ifs_true(node):
op = node.op
op = node.op
if not isinstance(op, IfElse):
return False
t_ins = node.inputs[1:][:op.n_outs]
replace = {}
for idx,tval in enumerate(t_ins):
for idx, tval in enumerate(t_ins):
if (tval.owner and isinstance(tval.owner.op, IfElse) and
tval.owner.inputs[0] == node.inputs[0]):
ins_op = tval.owner.op
ins_t = tval.owner.inputs[1:][:ins_op.n_outs]
replace[idx+1] = ins_t[tval.owner.outputs.index(tval)]
ins_t = tval.owner.inputs[1:][:ins_op.n_outs]
replace[idx + 1] = ins_t[tval.owner.outputs.index(tval)]
if len(replace.items()) == 0:
return False
old_ins = list(node.inputs)
for pos,var in replace.items():
for pos, var in replace.items():
old_ins[pos] = var
return op.make_node(*old_ins).outputs
......@@ -365,24 +354,26 @@ def cond_merge_ifs_false(node):
f_ins = node.inputs[1:][op.n_outs:]
replace = {}
for idx,fval in enumerate(f_ins):
for idx, fval in enumerate(f_ins):
if (fval.owner and isinstance(fval.owner.op, IfElse) and
fval.owner.inputs[0] == node.inputs[0]):
ins_op = fval.owner.op
ins_t = fval.owner.inputs[1:][ins_op.n_outs:]
replace[idx+1+op.n_outs] = ins_t[fval.owner.outputs.index(fval)]
ins_t = fval.owner.inputs[1:][ins_op.n_outs:]
replace[idx + 1 + op.n_outs] = \
ins_t[fval.owner.outputs.index(fval)]
if len(replace.items()) == 0:
return False
old_ins = list(node.inputs)
for pos,var in replace.items():
for pos, var in replace.items():
old_ins[pos] = var
return op.make_node(*old_ins).outputs
class CondMerge(gof.Optimizer):
""" Graph Optimizer that merges different cond ops """
def add_requirements(self,env):
def add_requirements(self, env):
env.extend(gof.toolbox.ReplaceValidate())
def apply(self, env):
......@@ -400,20 +391,20 @@ class CondMerge(gof.Optimizer):
pl_ts = proposal.inputs[1:][:proposal.op.n_outs]
pl_fs = proposal.inputs[1:][proposal.op.n_outs:]
new_ins = ([merging_node.inputs[0]] +
mn_ts + pl_ts + mn_fs + pl_fs )
mn_ts + pl_ts + mn_fs + pl_fs)
mn_name = '?'
if merging_node.op.name:
mn_name = merging_node.op.name
pl_name = '?'
mn_n_ts = len(mn_ts)
mn_n_fs = len(mn_fs)
mn_n_ts = len(mn_ts)
mn_n_fs = len(mn_fs)
if proposal.op.name:
pl_name = proposal.op.name
new_ifelse = IfElse(
n_outs = len(mn_ts+pl_ts),
n_outs=len(mn_ts + pl_ts),
as_view=False,
gpu = False,
name = mn_name+'&'+pl_name)
gpu=False,
name=mn_name + '&' + pl_name)
print 'here'
new_outs = new_ifelse.make_node(*new_ins).outputs
new_outs = [clone(x) for x in new_outs]
......@@ -443,7 +434,7 @@ def cond_remove_identical(node):
out_map = {}
for idx in xrange(len(node.outputs)):
if idx not in out_map:
for jdx in xrange(idx+1,len(node.outputs)):
for jdx in xrange(idx + 1, len(node.outputs)):
if (ts[idx] == ts[jdx] and
fs[idx] == fs[jdx] and
jdx not in out_map):
......@@ -454,7 +445,7 @@ def cond_remove_identical(node):
nw_ts = []
nw_fs = []
inv_map = {}
inv_map = {}
pos = 0
for idx in xrange(len(node.outputs)):
if idx not in out_map:
......@@ -463,12 +454,10 @@ def cond_remove_identical(node):
nw_ts.append(ts[idx])
nw_fs.append(fs[idx])
new_ifelse = IfElse(n_outs = len(nw_ts),
as_view = op.as_view,
gpu = op.gpu,
name = op.name)
new_ifelse = IfElse(n_outs=len(nw_ts),
as_view=op.as_view,
gpu=op.gpu,
name=op.name)
new_ins = [node.inputs[0]] + nw_ts + nw_fs
new_outs = new_ifelse.make_node(*new_ins).outputs
......@@ -476,14 +465,12 @@ def cond_remove_identical(node):
rval = []
for idx in xrange(len(node.outputs)):
if idx in out_map.keys():
rval += [new_outs[inv_map[out_map[idx]]] ]
rval += [new_outs[inv_map[out_map[idx]]]]
else:
rval += [new_outs[inv_map[idx]]]
return rval
acceptable_ops = (theano.tensor.basic.Dot,
theano.tensor.basic.Reshape,
theano.tensor.basic.Shape,
......@@ -494,7 +481,7 @@ acceptable_ops = (theano.tensor.basic.Dot,
theano.tensor.basic.Rebroadcast,
theano.tensor.basic.Alloc,
theano.tensor.elemwise.Elemwise,
theano.tensor.elemwise.DimShuffle )
theano.tensor.elemwise.DimShuffle)
@gof.local_optimizer([None])
......@@ -504,7 +491,8 @@ def cond_lift_single_if(main_node):
all_inp_nodes = set()
for inp in main_node.inputs:
all_inp_nodes.add(inp.owner)
ifnodes = [ x for x in list(all_inp_nodes) if x and isinstance(x.op, IfElse)]
ifnodes = [x for x in list(all_inp_nodes)
if x and isinstance(x.op, IfElse)]
# if we have multiple ifs as inputs .. it all becomes quite complicated
# :)
if len(ifnodes) != 1:
......@@ -528,7 +516,7 @@ def cond_lift_single_if(main_node):
else:
true_ins.append(x)
false_ins.append(x)
true_eval = mop.make_node(*true_ins).outputs
true_eval = mop.make_node(*true_ins).outputs
false_eval = mop.make_node(*false_ins).outputs
#true_eval = clone(outs, replace = dict(zip(node.outputs, ts)))
#false_eval = clone(outs, replace = dict(zip(node.outputs, fs)))
......@@ -539,7 +527,6 @@ def cond_lift_single_if(main_node):
return nw_outs
@gof.local_optimizer([None])
def cond_merge_random_op(main_node):
if isinstance(main_node.op, IfElse):
......@@ -548,7 +535,8 @@ def cond_merge_random_op(main_node):
all_inp_nodes = set()
for inp in main_node.inputs:
all_inp_nodes.add(inp.owner)
cond_nodes = [ x for x in list(all_inp_nodes) if x and isinstance(x.op, IfElse)]
cond_nodes = [x for x in list(all_inp_nodes)
if x and isinstance(x.op, IfElse)]
if len(cond_nodes) < 2:
return False
......@@ -564,20 +552,20 @@ def cond_merge_random_op(main_node):
pl_ts = proposal.inputs[1:][:proposal.op.n_outs]
pl_fs = proposal.inputs[1:][proposal.op.n_outs:]
new_ins = ([merging_node.inputs[0]] +
mn_ts + pl_ts + mn_fs + pl_fs )
mn_ts + pl_ts + mn_fs + pl_fs)
mn_name = '?'
if merging_node.op.name:
mn_name = merging_node.op.name
pl_name = '?'
mn_n_ts = len(mn_ts)
mn_n_fs = len(mn_fs)
mn_n_ts = len(mn_ts)
mn_n_fs = len(mn_fs)
if proposal.op.name:
pl_name = proposal.op.name
new_ifelse = IfElse(
n_outs = len(mn_ts+pl_ts),
n_outs=len(mn_ts + pl_ts),
as_view=False,
gpu = False,
name = mn_name+'&'+pl_name)
gpu=False,
name=mn_name + '&' + pl_name)
new_outs = new_ifelse.make_node(*new_ins).outputs
old_outs = []
if type(merging_node.outputs) not in (list, tuple):
......@@ -593,19 +581,16 @@ def cond_merge_random_op(main_node):
return main_outs
pushout_equilibrium = gof.EquilibriumDB()
pushout_equilibrium.register("ifelse_lift",
opt.in2out(cond_lift_single_if,
ignore_newtrees = True),
ignore_newtrees=True),
'fast_run', 'ifelse')
pushout_equilibrium.register("ifelse_merge_ifs",
opt.in2out(cond_merge_random_op,
ignore_newtrees = True),
ignore_newtrees=True),
'fast_run', 'ifelse')
......@@ -615,7 +600,7 @@ pushout_equilibrium.register("ifelse_merge_nodes",
pushout_equilibrium.register("ifelse_remove_identical_inside",
opt.in2out(cond_remove_identical,
ignore_newtrees = True),
ignore_newtrees=True),
'fast_run', 'ifelse')
pushout_equilibrium.register('ifelse_sameCondTrue_inside',
......@@ -644,8 +629,6 @@ ifelse_seqopt.register('ifelse_sameCondTrue',
3, 'fast_run', 'ifelse')
ifelse_seqopt.register('ifelse_sameCondFalse',
opt.in2out(cond_merge_ifs_false,
ignore_newtrees=True),
......@@ -654,6 +637,5 @@ ifelse_seqopt.register('ifelse_sameCondFalse',
ifelse_seqopt.register('ifelse_removeIdenetical',
opt.in2out(cond_remove_identical,
ignore_newtrees = True),
ignore_newtrees=True),
7, 'fast_run', 'ifelse')
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