explosion/spaCy
1
1
Fork 0
mirror of https://github.com/explosion/spaCy.git synced 2025-01-12 18:26:30 +03:00
Code Issues Packages Projects Releases Wiki Activity

* Move to fixing up ent_strings and dep_strings passing

Browse Source
This commit is contained in:
Matthew Honnibal 2015-03-14 11:09:55 -04:00
parent 31fad99518
commit 9061bbaf61
1 changed files with 8 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

20
bin/parser/train.py
View File

@ -213,10 +213,6 @@ def train(Language, train_loc, model_dir, n_iter=15, feat_set=u'basic', seed=0,
if n_sents > 0: if n_sents > 0:
gold_tuples = gold_tuples[:n_sents] gold_tuples = gold_tuples[:n_sents]
nlp = Language() nlp = Language()
ent_strings = [None] * (max(nlp.entity.moves.label_ids.values()) + 1)
for label, i in nlp.entity.moves.label_ids.items():
if i >= 0:
ent_strings[i] = label
print "Itn.\tUAS\tNER F.\tTag %" print "Itn.\tUAS\tNER F.\tTag %"
for itn in range(n_iter): for itn in range(n_iter):
@ -229,12 +225,11 @@ def train(Language, train_loc, model_dir, n_iter=15, feat_set=u'basic', seed=0,
for tokens in sents: for tokens in sents:
gold = GoldParse(tokens, annot_tuples) gold = GoldParse(tokens, annot_tuples)
nlp.tagger(tokens) nlp.tagger(tokens)
nlp.entity.train(tokens, gold, force_gold=force_gold) nlp.parser.train(tokens, gold, force_gold=force_gold)
#nlp.parser.train(tokens, gold, force_gold=force_gold) #nlp.entity.train(tokens, gold, force_gold=force_gold)
nlp.tagger.train(tokens, gold.tags) nlp.tagger.train(tokens, gold.tags)
nlp.entity(tokens) #nlp.entity(tokens)
tokens._ent_strings = tuple(ent_strings)
nlp.parser(tokens) nlp.parser(tokens)
scorer.score(tokens, gold, verbose=False) scorer.score(tokens, gold, verbose=False)
print '%d:\t%.3f\t%.3f\t%.3f' % (itn, scorer.uas, scorer.ents_f, scorer.tags_acc) print '%d:\t%.3f\t%.3f\t%.3f' % (itn, scorer.uas, scorer.ents_f, scorer.tags_acc)
@ -244,7 +239,7 @@ def train(Language, train_loc, model_dir, n_iter=15, feat_set=u'basic', seed=0,
nlp.tagger.model.end_training() nlp.tagger.model.end_training()
def evaluate(Language, dev_loc, model_dir, gold_preproc=False, verbose=False): def evaluate(Language, dev_loc, model_dir, gold_preproc=False, verbose=True):
assert not gold_preproc assert not gold_preproc
nlp = Language() nlp = Language()
gold_tuples = read_docparse_file(dev_loc) gold_tuples = read_docparse_file(dev_loc)
@ -260,12 +255,13 @@ def evaluate(Language, dev_loc, model_dir, gold_preproc=False, verbose=False):
train_loc=("Training file location",), train_loc=("Training file location",),
dev_loc=("Dev. file location",), dev_loc=("Dev. file location",),
model_dir=("Location of output model directory",), model_dir=("Location of output model directory",),
n_sents=("Number of training sentences", "option", "n", int) n_sents=("Number of training sentences", "option", "n", int),
verbose=("Verbose error reporting", "flag", "v", bool),
) )
def main(train_loc, dev_loc, model_dir, n_sents=0): def main(train_loc, dev_loc, model_dir, n_sents=0, verbose=False):
train(English, train_loc, model_dir, train(English, train_loc, model_dir,
gold_preproc=False, force_gold=False, n_sents=n_sents) gold_preproc=False, force_gold=False, n_sents=n_sents)
scorer = evaluate(English, dev_loc, model_dir, gold_preproc=False, verbose=False) scorer = evaluate(English, dev_loc, model_dir, gold_preproc=False, verbose=verbose)
print 'POS', scorer.tags_acc print 'POS', scorer.tags_acc
print 'UAS', scorer.uas print 'UAS', scorer.uas
print 'LAS', scorer.las print 'LAS', scorer.las