explosion/spaCy
1
1
Fork 0
mirror of https://github.com/explosion/spaCy.git synced 2025-07-13 17:52:31 +03:00
Code Issues Packages Projects Releases Wiki Activity

Add parameter for gradient noise

Browse Source
This commit is contained in:
Matthew Honnibal 2016-08-05 18:24:01 +02:00
parent a664aa8180
commit 532fa36c13
1 changed files with 7 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
bin/parser/conll_train.py
View File

@ -119,7 +119,7 @@ def score_sents(nlp, gold_tuples):
def train(Language, gold_tuples, model_dir, dev_loc, n_iter=15, feat_set=u'basic', def train(Language, gold_tuples, model_dir, dev_loc, n_iter=15, feat_set=u'basic',
learn_rate=0.001, update_step='sgd_cm', learn_rate=0.001, noise=0.01, update_step='sgd_cm',
batch_norm=False, seed=0, gold_preproc=False, force_gold=False): batch_norm=False, seed=0, gold_preproc=False, force_gold=False):
dep_model_dir = path.join(model_dir, 'deps') dep_model_dir = path.join(model_dir, 'deps')
pos_model_dir = path.join(model_dir, 'pos') pos_model_dir = path.join(model_dir, 'pos')
@ -147,6 +147,7 @@ def train(Language, gold_tuples, model_dir, dev_loc, n_iter=15, feat_set=u'basic
batch_norm=batch_norm, batch_norm=batch_norm,
eta=learn_rate, eta=learn_rate,
mu=0.9, mu=0.9,
noise=noise,
ensemble_size=1, ensemble_size=1,
rho=rho) rho=rho)
@ -171,7 +172,7 @@ def train(Language, gold_tuples, model_dir, dev_loc, n_iter=15, feat_set=u'basic
except KeyboardInterrupt: except KeyboardInterrupt:
print("Saving model...") print("Saving model...")
break break
nlp.end_training(model_dir) nlp.parser.model.end_training()
print("Saved. Evaluating...") print("Saved. Evaluating...")
return nlp return nlp
@ -208,10 +209,11 @@ def _train_epoch(nlp, gold_tuples, eg_seen, itn, dev_loc, micro_eval):
batch_norm=("Use batch normalization and residual connections", "flag", "b"), batch_norm=("Use batch normalization and residual connections", "flag", "b"),
update_step=("Update step", "option", "u", str), update_step=("Update step", "option", "u", str),
learn_rate=("Learn rate", "option", "e", float), learn_rate=("Learn rate", "option", "e", float),
gradient_noise=("Gradient noise", "option", "w", float),
neural=("Use neural network?", "flag", "N") neural=("Use neural network?", "flag", "N")
) )
def main(train_loc, dev_loc, model_dir, n_iter=15, neural=False, batch_norm=False, def main(train_loc, dev_loc, model_dir, n_iter=15, neural=False, batch_norm=False,
learn_rate=0.001, update_step='sgd_cm'): learn_rate=0.001, gradient_noise=0.1, update_step='sgd_cm'):
with io.open(train_loc, 'r', encoding='utf8') as file_: with io.open(train_loc, 'r', encoding='utf8') as file_:
train_sents = list(read_conll(file_)) train_sents = list(read_conll(file_))
# Preprocess training data here before ArcEager.get_labels() is called # Preprocess training data here before ArcEager.get_labels() is called
@ -221,7 +223,8 @@ def main(train_loc, dev_loc, model_dir, n_iter=15, neural=False, batch_norm=Fals
feat_set='neural' if neural else 'basic', feat_set='neural' if neural else 'basic',
batch_norm=batch_norm, batch_norm=batch_norm,
learn_rate=learn_rate, learn_rate=learn_rate,
update_step=update_step) update_step=update_step,
noise=gradient_noise)
scorer = score_file(nlp, dev_loc) scorer = score_file(nlp, dev_loc)
print('TOK', scorer.token_acc) print('TOK', scorer.token_acc)