Skip to content

P
pytensor
提交 1c974f4f authored 作者: Arnaud Bergeron's avatar Arnaud Bergeron
浏览文件
操作

- Only run local optimizers on the ops they are registered for.

- Fix existing optimizers in base code to register properly.
上级 ab660ca3
隐藏空白字符变更
内嵌 并排
正在显示 包含 90 行增加 和 59 行删除
theano/gof/opt.py
浏览文件 @ 1c974f4f
......@@ -736,6 +736,14 @@ class LocalOptimizer(object):
_optimizer_idx[0] += 1
return self._optimizer_idx
def tracks(self):
"""
Return the list of op classes that this opt applies to.
Return None to apply to all nodes.
"""
return None
def transform(self, node):
"""Transform a subgraph whose output is `node`.
......@@ -791,9 +799,15 @@ class FromFunctionLocalOptimizer(LocalOptimizer):
id(self))
def local_optimizer(*tracks):
def local_optimizer(tracks):
def decorator(f):
"""WRITEME"""
if tracks is not None:
if len(tracks) is 0:
raise ValueError, ("Use None instead of an empty list to apply to all nodes.", f.__module__, f.__name__)
for t in tracks:
if not (isinstance(t, type) or isinstance(t, op.Op)):
raise ValueError, ("Tracks are op classes or instances", f.__module__, f.__name__)
rval = FromFunctionLocalOptimizer(f, tracks)
rval.__name__ = f.__name__
return rval
......@@ -870,7 +884,7 @@ class OpSub(LocalOptimizer):
return self.op1
def tracks(self):
return [[self.op1]]
return [self.op1]
def transform(self, node):
if node.op != self.op1:
......@@ -901,7 +915,7 @@ class OpRemove(LocalOptimizer):
return self.op
def tracks(self):
return [[self.op]]
return [self.op]
def transform(self, node):
if node.op != self.op:
......@@ -1500,12 +1514,17 @@ class EquilibriumOptimizer(NavigatorOptimizer):
None,
ignore_newtrees=True,
failure_callback=failure_callback)
self.local_optimizers = []
self.local_optimizers_map = dict()
self.local_optimizers_all = []
self.global_optimizers = []
for opt in optimizers:
if isinstance(opt, LocalOptimizer):
self.local_optimizers.append(opt)
if opt.tracks is None:
self.local_optimizers_all.append(opt)
else:
for c in opt.tracks():
self.local_optimizers_map.setdefault(c, []).append(opt)
else:
self.global_optimizers.append(opt)
self.max_depth = max_depth
......@@ -1513,10 +1532,21 @@ class EquilibriumOptimizer(NavigatorOptimizer):
assert self.max_use_ratio is not None, (
'max_use_ratio has to be a number')
def get_local_optimizers(self):
for opt in self.local_optimizers_all:
yield opt
# if repeat is not a problem we can drop the set
s = set()
for lopt in self.local_optimizers_map.values():
for opt in lopt:
if opt not in s:
yield opt
s.add(opt)
def add_requirements(self, fgraph):
super(EquilibriumOptimizer, self).add_requirements(fgraph)
fgraph.attach_feature(ChangeTracker())
for opt in self.local_optimizers:
for opt in self.get_local_optimizers():
opt.add_requirements(fgraph)
for opt in self.global_optimizers:
opt.add_requirements(fgraph)
......@@ -1542,7 +1572,7 @@ class EquilibriumOptimizer(NavigatorOptimizer):
time_opts = {}
io_toposort_timing = []
nb_nodes = []
for opt in self.global_optimizers + self.local_optimizers:
for opt in self.global_optimizers + list(self.get_local_optimizers()):
global_process_count.setdefault(opt, 0)
time_opts.setdefault(opt, 0)
......@@ -1595,7 +1625,9 @@ class EquilibriumOptimizer(NavigatorOptimizer):
node = q.pop()
current_node = node
for lopt in self.local_optimizers:
for lopt in (self.local_optimizers_all +
self.local_optimizers_map.get(type(node.op), []) +
self.local_optimizers_map.get(node.op, [])):
t_opt = time.time()
lopt_change = self.process_node(fgraph, node, lopt)
time_opts[lopt] += time.time() - t_opt
......@@ -1634,7 +1666,7 @@ class EquilibriumOptimizer(NavigatorOptimizer):
print >> stream, "%s%s %s id=%i" % (
(' ' * level), self.__class__.__name__, name, id(self))
if depth != 0:
for lopt in self.local_optimizers:
for lopt in self.get_local_optimizers():
lopt.print_summary(stream, level=(level + 2),
depth=(depth - 1))
......@@ -1654,7 +1686,7 @@ class EquilibriumOptimizer(NavigatorOptimizer):
start_nb_nodes, end_nb_nodes, max_nb_nodes)
print >> stream, blanc, " time io_toposort %.3fs" % sum(
io_toposort_timing)
s = sum([time_opts[o] for o in opt.local_optimizers])
s = sum([time_opts[o] for o in opt.get_local_optimizers()])
print >> stream, blanc, " time in local optimizers %.3fs" % s
s = sum([time_opts[o] for o in opt.global_optimizers])
print >> stream, blanc, " time in global optimizers %.3fs" % s
......@@ -1679,7 +1711,7 @@ class EquilibriumOptimizer(NavigatorOptimizer):
not_used = 0
not_used_time = 0
process_count = {}
for o in opt.global_optimizers + opt.local_optimizers:
for o in opt.global_optimizers + opt.get_local_optimizers():
process_count.setdefault(o, 0)
for count in loop_process_count:
for o, v in count.iteritems():
......@@ -1707,8 +1739,8 @@ class EquilibriumOptimizer(NavigatorOptimizer):
#(opt, loop_timing, loop_process_count, max_nb_nodes,
# global_opt_timing, nb_nodes, time_opts, io_toposort_timing) = prof1
local_optimizers = set(prof1[0].local_optimizers).union(
prof2[0].local_optimizers)
local_optimizers = set(prof1[0].get_local_optimizers()).union(
prof2[0].get_local_optimizers())
global_optimizers = set(prof1[0].global_optimizers).union(
prof2[0].global_optimizers)
new_opt = EquilibriumOptimizer(
......
theano/scan_module/scan_opt.py
浏览文件 @ 1c974f4f
......@@ -49,7 +49,7 @@ def info(*msg):
_logger.info('INFO theano.scan: ' + ' '.join(msg))
@gof.local_optimizer([None])
@gof.local_optimizer([scan_op.Scan])
def remove_constants_and_unused_inputs_scan(node):
'''
Move constants into the inner graph, and remove unused inputs.
......@@ -1337,7 +1337,7 @@ def make_equiv(lo, li):
return left, right
@gof.local_optimizer([None])
@gof.local_optimizer([scan_op.Scan])
def scan_merge_inouts(node):
if not isinstance(node.op, scan_op.Scan):
return False
......
theano/tensor/blas.py
浏览文件 @ 1c974f4f
......@@ -1645,7 +1645,7 @@ class Dot22(GemmRelated):
_dot22 = Dot22()
@local_optimizer([T._dot])
@local_optimizer([T.Dot])
def local_dot_to_dot22(node):
# This works for tensor.outer too because basic.outer is a macro that
# produces a dot(dimshuffle,dimshuffle) of form 4 below
......@@ -2025,7 +2025,7 @@ blas_optdb.register('local_dot22_to_dot22scalar',
#from opt import register_specialize, register_canonicalize
#@register_specialize
@local_optimizer([])
@local_optimizer([T.sub, T.add])
def local_print_as_we_go_along(node):
if node.op in (T.sub, T.add):
debugprint(node)
theano/tensor/nnet/nnet.py
浏览文件 @ 1c974f4f
......@@ -589,7 +589,7 @@ opt.local_mul_canonizer.add_simplifier(softmax_simplifier,
if 0:
@opt.register_specialize
@gof.local_optimizer([])
@gof.local_optimizer([tensor.add])
def local_softmax_grad(node):
'''dy*sm - DimShuffle{0,'x'}(sum{1}(dy*sm))*sm -> softmax_grad(dy,sm)'''
#TODO what if the signs are changed?
......@@ -1417,7 +1417,7 @@ def _is_const(z, val, approx=False):
@opt.register_specialize
@gof.local_optimizer([])
@gof.local_optimizer([subtensor.AdvancedSubtensor])
def local_advanced_indexing_crossentropy_onehot(node):
log = None
sm = None
......
theano/tensor/opt.py
浏览文件 @ 1c974f4f
......@@ -347,7 +347,7 @@ compile.optdb['canonicalize'].register(
@register_canonicalize
@register_stabilize
@gof.local_optimizer([None])
@gof.local_optimizer([T.Dot])
def local_0_dot_x(node):
if not isinstance(node.op, T.Dot):
return False
......@@ -390,7 +390,7 @@ def local_0_dot_x(node):
######################
@gof.local_optimizer([None, None])
@gof.local_optimizer([DimShuffle])
def local_dimshuffle_lift(node):
"""
"Lifts" DimShuffle through Elemwise operations and merges
......@@ -431,7 +431,7 @@ def local_dimshuffle_lift(node):
@register_canonicalize
@gof.local_optimizer([])
@gof.local_optimizer([T.DimShuffle])
def local_lift_transpose_through_dot(node):
"""
dot(x,y).T -> dot(y.T, x.T)
......@@ -456,7 +456,7 @@ def local_lift_transpose_through_dot(node):
return [T.dot(y.T, x.T)]
@gof.local_optimizer([])
@gof.local_optimizer([DimShuffle])
def dimshuffle_as_view(node):
op = node.op
if not isinstance(op, DimShuffle) or op.inplace:
......@@ -476,7 +476,7 @@ register_specialize(local_dimshuffle_lift)
@register_canonicalize
@gof.local_optimizer([])
@gof.local_optimizer([T.DimShuffle])
def local_dimshuffle_no_inplace_at_canonicalize(node):
if isinstance(node.op, T.DimShuffle) and node.op.inplace:
return [T.DimShuffle(node.op.input_broadcastable,
......@@ -1213,9 +1213,10 @@ def local_shape_to_shape_i(node):
return [shape_feature.make_vector_shape(node.inputs[0])]
# TODO: Not sure what type of node we are expecting here
@register_specialize
@register_canonicalize
@gof.local_optimizer([T._shape])
@gof.local_optimizer(None)
def local_track_shape_i(node):
try:
shape_feature = node.fgraph.shape_feature
......@@ -1415,7 +1416,7 @@ def local_remove_useless_assert(node):
return [assert_(node.inputs[0], *cond)]
@gof.local_optimizer([T.Alloc])
@gof.local_optimizer([T.Elemwise])
def local_alloc_elemwise(node):
"""
elemwise(alloc(x, shp), ..., y.TensorType(BROADCAST CONDITION))
......@@ -1534,7 +1535,7 @@ else:
@register_canonicalize
@gof.local_optimizer([])
@gof.local_optimizer([T.Elemwise])
def local_upcast_elemwise_constant_inputs(node):
"""This explicitly upcasts constant inputs to elemwise Ops, when
those Ops do implicit upcasting anyway.
......@@ -1682,7 +1683,7 @@ def local_useless_subtensor(node):
@register_canonicalize
@gof.local_optimizer([])
@gof.local_optimizer([Subtensor])
def local_subtensor_lift(node):
"""
unary(x)[idx] -> unary(x[idx])#any broadcast pattern.
......@@ -1892,7 +1893,7 @@ def merge_two_slices(slice1, len1, slice2, len2):
@register_canonicalize
@register_specialize
@gof.local_optimizer([])
@gof.local_optimizer([Subtensor])
def local_subtensor_merge(node):
"""
Refactored optimization to deal with all cases of tensor merging.
......@@ -1954,7 +1955,7 @@ def local_subtensor_merge(node):
@register_canonicalize
@register_specialize
@gof.local_optimizer([])
@gof.local_optimizer([Subtensor])
def local_subtensor_of_alloc(node):
"""alloc[x:y] -> alloc"""
if not isinstance(node.op, Subtensor):
......@@ -2007,7 +2008,7 @@ def local_subtensor_of_alloc(node):
@register_canonicalize
@gof.local_optimizer([None])
@gof.local_optimizer([T.add])
def local_IncSubtensor_serialize(node):
"""
When using Subtensor, gradient graphs can be ugly.
......@@ -2079,7 +2080,7 @@ compile.optdb.register('pre_local_IncSubtensor_serialize',
#after priority 50 Destructive inplace operations
#gemm is the first one now, at priority 70
@gof.local_optimizer([None])
@gof.local_optimizer([IncSubtensor]) # XXX: GPU
def local_inplace_setsubtensor(node):
"""
Also work for GpuIncSubtensor
......@@ -2098,7 +2099,7 @@ compile.optdb.register('local_inplace_setsubtensor',
'fast_run', 'inplace') # DEBUG
@gof.local_optimizer([None])
@gof.local_optimizer([AdvancedIncSubtensor1]) # XXX: GPU
def local_inplace_incsubtensor1(node):
""" also work for GpuAdvancedIncSubtensor1 """
if isinstance(node.op, AdvancedIncSubtensor1) and not node.op.inplace:
......@@ -2116,7 +2117,7 @@ compile.optdb.register('local_inplace_incsubtensor1',
@register_canonicalize
@register_stabilize
@gof.local_optimizer([None])
@gof.local_optimizer([IncSubtensor])
def local_incsubtensor_of_allocs(node):
"""
IncSubtensor(x, zeros, idx) -> x
......@@ -2139,7 +2140,7 @@ def local_incsubtensor_of_allocs(node):
@register_canonicalize
@register_stabilize
@gof.local_optimizer([None])
@gof.local_optimizer([IncSubtensor])
def local_setsubtensor_of_allocs(node):
"""
SetSubtensor(x, x[idx], idx) -> x
......@@ -2286,7 +2287,7 @@ def local_join_1(node):
###############
@register_canonicalize
@gof.local_optimizer([])
@gof.local_optimizer([T.Elemwise])
def local_remove_switch_const_cond(node):
"""
This optimization makes the following changes in the graph:
......@@ -2369,7 +2370,7 @@ def local_mul_switch_sink(node):
@register_canonicalize
@gof.local_optimizer([T.true_div])
@gof.local_optimizer([T.true_div, T.int_div, T.floor_div])
def local_div_switch_sink(node):
"""
This optimization makes the folowing changes in the graph:
......@@ -2413,7 +2414,7 @@ def local_div_switch_sink(node):
################
@register_canonicalize
@register_stabilize
@gof.local_optimizer([])
@gof.local_optimizer([T.Flatten])
def local_flatten_lift(node):
"""
Flatten(UnaryElemwise(x)) -> UnaryElemwise(Flatten(x))
......@@ -2434,7 +2435,7 @@ def local_flatten_lift(node):
##################
@gof.local_optimizer([None, None])
@gof.local_optimizer([T.Reshape])
def local_reshape_chain(node):
"""
Reshape(Reshape(shape1),shape2) -> Reshape(shape2)
......@@ -2462,7 +2463,7 @@ register_canonicalize(local_reshape_chain)
@register_canonicalize
@register_stabilize
@gof.local_optimizer([])
@gof.local_optimizer([T.Reshape])
def local_reshape_lift(node):
"""
Reshape(UnaryElemwise(x)) -> UnaryElemwise(Reshape(x))
......@@ -2482,7 +2483,7 @@ def local_reshape_lift(node):
if 0:
# TODO: Test that this optimziation works.
@register_canonicalize
@gof.local_optimizer([])
@gof.local_optimizer([T.Reshape])
def local_scalar_reshape(node):
"""Eliminate reshape Ops whose inputs and outputs are scalars """
if isinstance(node.op, T.Reshape):
......@@ -2498,7 +2499,7 @@ if 0:
# TODO: Remember to take into account the new sum dtype argument if this
# optimization is enabled.
@register_canonicalize
@gof.local_optimizer([])
@gof.local_optimizer([T.Sum])
def local_sum_over_empty(node):
if isinstance(node.op, T.Sum):
# This optimization needs ShapeOpt and fgraph.shape_feature
......@@ -2520,7 +2521,7 @@ if 0:
##################
@gof.local_optimizer([None, T.fill])
@gof.local_optimizer([T.Elemwise])
def local_fill_cut(node):
"""
f(fill(a,b), c) -> f(b, c)
......@@ -2574,7 +2575,7 @@ register_canonicalize(local_fill_cut)
register_canonicalize(gof.OpRemove(T.tensor_copy), name='remove_tensor_copy')
@gof.local_optimizer([None, T.fill])
@gof.local_optimizer([T.Elemwise])
def local_fill_sink(node):
"""
f(fill(a, b), fill(c, d), e) -> fill(a, fill(c, f(b, d, e)))
......@@ -2662,8 +2663,7 @@ class Canonizer(gof.LocalOptimizer):
self.external_simplifiers.append((reason, simplifier))
def tracks(self):
return [[self.main, None], [self.inverse, None],
[self.reciprocal, None]]
return [self.main, self.inverse, self.reciprocal]
def get_num_denum(self, input):
"""
......@@ -3051,7 +3051,7 @@ register_canonicalize(local_neg_to_mul)
@register_specialize
@gof.local_optimizer([])
@gof.local_optimizer([T.Sum])
def local_sum_mul_by_scalar(node):
"""sum(scalar * smth) -> scalar * sum(smth)
sum(-smth) -> -sum(smth)
......@@ -3088,7 +3088,7 @@ def local_sum_mul_by_scalar(node):
@register_specialize
@gof.local_optimizer([])
@gof.local_optimizer([T.Elemwise])
def local_elemwise_sub_zeros(node):
"""
Elemwise{sub}(X,X) -> zeros_like(X)
......@@ -3102,7 +3102,7 @@ def local_elemwise_sub_zeros(node):
@register_canonicalize
@register_specialize
@gof.local_optimizer([])
@gof.local_optimizer([T.Sum])
def local_sum_div_dimshuffle(node):
'''sum(a / dimshuffle{...}(b), axis=l) -> sum(a, axis={...}) / b,
if dimension l of the DimShuffle is 'x'.'''
......@@ -3191,7 +3191,7 @@ def local_sum_div_dimshuffle(node):
@register_canonicalize
@gof.local_optimizer([])
@gof.local_optimizer([T.Sum])
def local_sum_all_to_none(node):
"""Sum{0,1,...N} -> Sum{}"""
if isinstance(node.op, T.Sum):
......@@ -3204,7 +3204,7 @@ def local_sum_all_to_none(node):
@register_canonicalize
@gof.local_optimizer([])
@gof.local_optimizer([T.Sum])
def local_sum_sum(node):
"""
Sum(Sum()) -> Sum
......@@ -3272,7 +3272,7 @@ def local_sum_sum(node):
@register_canonicalize
@gof.local_optimizer([])
@gof.local_optimizer([T.CAReduce])
def local_cut_useless_reduce(node):
"""Sum(a, axis=[]) -> a """
if isinstance(node.op, T.CAReduce):
......@@ -3288,7 +3288,7 @@ def local_cut_useless_reduce(node):
#
#@register_canonicalize
@register_specialize
@gof.local_optimizer([])
@gof.local_optimizer([T.CAReduce])
def local_reduce_broadcastable(node):
"""Remove reduction over broadcastable dimensions"""
if isinstance(node.op, T.CAReduce):
......@@ -3327,7 +3327,7 @@ def local_reduce_broadcastable(node):
@register_specialize
@gof.local_optimizer([])
@gof.local_optimizer([T.Sum])
def local_sum_alloc(node):
""" sum(alloc(constant,shapes...)) => constant*prod(shapes)"""
if isinstance(node.op, T.Sum):
......@@ -3734,7 +3734,7 @@ def local_abs_lift(node):
@register_specialize
@gof.local_optimizer([])
@gof.local_optimizer([T.mul])
def local_abs_merge(node):
"""
merge abs generated by local_abs_lift when the canonizer don't
......@@ -3909,8 +3909,7 @@ def attempt_distribution(factor, num, denum):
neg_pairs))), num, denum
@gof.local_optimizer([T.mul, T.add, T.mul], [T.mul, T.sub, T.mul],
[T.mul, T.add, T.true_div], [T.mul, T.sub, T.true_div])
@gof.local_optimizer([T.mul])
def local_greedy_distributor(node):
"""
This optimization tries to apply distributivity of multiplication
......@@ -3976,7 +3975,7 @@ register_canonicalize(local_greedy_distributor)
register_stabilize(local_greedy_distributor)
@gof.local_optimizer([None])
@gof.local_optimizer(None)
def constant_folding(node):
for input in node.inputs:
if not isinstance(input, Constant):
......
theano/tensor/raw_random.py
浏览文件 @ 1c974f4f
......@@ -816,7 +816,7 @@ def multinomial(random_state, size=None, n=1, pvals=[0.5, 0.5],
return op(random_state, size, n, pvals)
@gof.local_optimizer([None])
@gof.local_optimizer([RandomFunction])
def random_make_inplace(node):
op = node.op
if isinstance(op, RandomFunction) and not op.inplace:
......
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