def do_fold(train_loc, model_loc, test_loc, out_loc, iters=5):
print model_loc, out_loc
tagger = GreedyTagger(str(model_loc), clean=True, reuse_idx=True)
train_strs = train_loc.open().read().strip().split('\n\n')
train = read_conll('\n\n'.join(train_strs))
tagger.train(train, nr_iter=iters)
tagger.save()
test_data = test_loc.open().read()
to_tag = read_conll(test_data)
tagger.add_tags(to_tag)
to_tag.write_tags(open(str(out_loc), 'w'))
def do_fold(train_loc, model_loc, test_loc, out_loc, iters=5):
print model_loc, out_loc
tagger = GreedyTagger(str(model_loc), clean=True, reuse_idx=True)
train_strs = train_loc.open().read().strip().split('\n\n')
train = read_conll('\n\n'.join(train_strs))
tagger.train(train, nr_iter=iters)
tagger.save()
test_data = test_loc.open().read()
to_tag = read_conll(test_data)
tagger.add_tags(to_tag)
to_tag.write_tags(open(str(out_loc), 'w'))
def main(train_loc, out_loc, label_set="MALT", allow_reattach=False, allow_moves=False):
parser_dir = Path('/tmp').join('parser')
if not parser_dir.exists():
parser_dir.mkdir()
grammar_loc = Path(train_loc).parent().join('rgrammar') if allow_reattach else None
parser = Parser(str(parser_dir), clean=True, label_set=label_set,
allow_reattach=allow_reattach, allow_move=allow_moves,
grammar_loc=grammar_loc)
train = read_conll(open(train_loc).read())
parser.add_gold_moves(train)
with open(out_loc, 'w') as out_file:
train.write_moves(out_file)
def main(model_loc, dev_loc):
parser = load_parser(model_loc)
sents = read_conll(open(dev_loc).read())
get_edit_stats(parser, sents)
def main(model_loc, dev_loc):
parser = load_parser(model_loc)
sents = read_conll(open(dev_loc).read())
get_edit_stats(parser, sents)