explosion/spaCy
1
1
Fork 0
mirror of https://github.com/explosion/spaCy.git synced 2024-12-25 17:36:30 +03:00
Code Issues Packages Projects Releases Wiki Activity

Use nogil in parser, and set L1 to 0.0 by default

Browse Source
This commit is contained in:
Matthew Honnibal 2017-03-15 09:31:01 -05:00
parent c61c501406
commit d719f8e77e
2 changed files with 20 additions and 17 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
spacy/syntax/parser.pxd
View File

@ -21,4 +21,4 @@ cdef class Parser:
cdef readonly TransitionSystem moves
cdef readonly object cfg
cdef int parseC(self, TokenC* tokens, int length, int nr_feat, int nr_class) with gil
cdef int parseC(self, TokenC* tokens, int length, int nr_feat) nogil

35
spacy/syntax/parser.pyx
View File

@ -84,15 +84,17 @@ cdef class ParserModel(AveragedPerceptron):
self.time += 1
best = arg_max_if_gold(eg.c.scores, eg.c.costs, eg.c.nr_class)
guess = eg.guess
cdef weight_t loss = 0.0
if guess == best:
return loss
for clas in [guess, best]:
loss += (-eg.c.costs[clas] - eg.c.scores[clas]) ** 2
d_loss = eg.c.scores[clas] - -eg.c.costs[clas]
for feat in eg.c.features[:eg.c.nr_feat]:
self.update_weight_ftrl(feat.key, clas, feat.value * d_loss)
return loss
if guess == best or best == -1:
return 0.0
for feat in eg.c.features[:eg.c.nr_feat]:
self.update_weight(feat.key, guess, feat.value * eg.c.costs[guess])
self.update_weight(feat.key, best, -feat.value * eg.c.costs[guess])
#for clas in [guess, best]:
# loss += (-eg.c.costs[clas] - eg.c.scores[clas]) ** 2
# d_loss = eg.c.scores[clas] - -eg.c.costs[clas]
# for feat in eg.c.features[:eg.c.nr_feat]:
# self.update_weight_ftrl(feat.key, clas, feat.value * d_loss)
return eg.c.costs[guess]
def update_from_histories(self, TransitionSystem moves, Doc doc, histories, weight_t min_grad=0.0):
cdef Pool mem = Pool()
@ -175,7 +177,7 @@ cdef class Parser:
elif 'features' not in cfg:
cfg['features'] = self.feature_templates
self.model = ParserModel(cfg['features'])
self.model.l1_penalty = cfg.get('L1', 1e-8)
self.model.l1_penalty = cfg.get('L1', 0.0)
self.model.learn_rate = cfg.get('learn_rate', 0.001)
self.cfg = cfg
@ -193,7 +195,7 @@ cdef class Parser:
"""
cdef int nr_feat = self.model.nr_feat
with nogil:
status = self.parseC(tokens.c, tokens.length, nr_feat, self.moves.n_moves)
status = self.parseC(tokens.c, tokens.length, nr_feat)
# Check for KeyboardInterrupt etc. Untested
PyErr_CheckSignals()
if status != 0:
@ -226,7 +228,7 @@ cdef class Parser:
if len(queue) == batch_size:
with nogil:
for i in cython.parallel.prange(batch_size, num_threads=n_threads):
status = self.parseC(doc_ptr[i], lengths[i], nr_feat, self.moves.n_moves)
status = self.parseC(doc_ptr[i], lengths[i], nr_feat)
if status != 0:
with gil:
raise ParserStateError(queue[i])
@ -238,7 +240,7 @@ cdef class Parser:
batch_size = len(queue)
with nogil:
for i in cython.parallel.prange(batch_size, num_threads=n_threads):
status = self.parseC(doc_ptr[i], lengths[i], nr_feat, self.moves.n_moves)
status = self.parseC(doc_ptr[i], lengths[i], nr_feat)
if status != 0:
with gil:
raise ParserStateError(queue[i])
@ -247,10 +249,11 @@ cdef class Parser:
self.moves.finalize_doc(doc)
yield doc
cdef int parseC(self, TokenC* tokens, int length, int nr_feat, int nr_class) with gil:
cdef int parseC(self, TokenC* tokens, int length, int nr_feat) nogil:
state = new StateC(tokens, length)
# NB: This can change self.moves.n_moves!
self.moves.initialize_state(state)
nr_class = self.moves.n_moves
cdef ExampleC eg
eg.nr_feat = nr_feat
@ -267,10 +270,10 @@ cdef class Parser:
self.model.set_scoresC(eg.scores, eg.features, eg.nr_feat)
guess = VecVec.arg_max_if_true(eg.scores, eg.is_valid, eg.nr_class)
if guess < 0:
return 1
action = self.moves.c[guess]
if not eg.is_valid[guess]:
return 1
action.do(state, action.label)
memset(eg.scores, 0, sizeof(eg.scores[0]) * eg.nr_class)