spaCy/examples/training/train_parser.py

99 lines
3.2 KiB
Python
Raw Normal View History

2017-10-26 16:15:37 +03:00
#!/usr/bin/env python
# coding: utf8
2017-11-01 02:43:22 +03:00
"""Example of training spaCy dependency parser, starting off with an existing
model or a blank model. For more details, see the documentation:
2017-11-07 14:00:43 +03:00
* Training: https://spacy.io/usage/training
* Dependency Parse: https://spacy.io/usage/linguistic-features#dependency-parse
2017-10-26 16:15:37 +03:00
2017-11-07 03:22:30 +03:00
Compatible with: spaCy v2.0.0+
2017-10-26 16:15:37 +03:00
"""
2016-10-16 18:05:55 +03:00
from __future__ import unicode_literals, print_function
2017-10-26 16:15:37 +03:00
2017-10-26 17:11:20 +03:00
import plac
2016-10-16 18:05:55 +03:00
import random
2017-10-26 16:15:37 +03:00
from pathlib import Path
2016-10-16 18:05:55 +03:00
import spacy
2017-10-26 16:15:37 +03:00
# training data
TRAIN_DATA = [
("They trade mortgage-backed securities.", {
'heads': [1, 1, 4, 4, 5, 1, 1],
'deps': ['nsubj', 'ROOT', 'compound', 'punct', 'nmod', 'dobj', 'punct']
}),
2017-12-13 12:51:05 +03:00
("I like London and Berlin.", {
'heads': [1, 1, 1, 2, 2, 1],
'deps': ['nsubj', 'ROOT', 'dobj', 'cc', 'conj', 'punct']
})
2017-10-26 16:15:37 +03:00
]
2017-10-26 17:11:20 +03:00
@plac.annotations(
model=("Model name. Defaults to blank 'en' model.", "option", "m", str),
output_dir=("Optional output directory", "option", "o", Path),
n_iter=("Number of training iterations", "option", "n", int))
def main(model=None, output_dir=None, n_iter=10):
2017-10-26 17:11:20 +03:00
"""Load the model, set up the pipeline and train the parser."""
2017-10-26 16:15:37 +03:00
if model is not None:
nlp = spacy.load(model) # load existing spaCy model
print("Loaded model '%s'" % model)
else:
nlp = spacy.blank('en') # create blank Language class
print("Created blank 'en' model")
# add the parser to the pipeline if it doesn't exist
# nlp.create_pipe works for built-ins that are registered with spaCy
if 'parser' not in nlp.pipe_names:
parser = nlp.create_pipe('parser')
nlp.add_pipe(parser, first=True)
# otherwise, get it, so we can add labels to it
else:
parser = nlp.get_pipe('parser')
# add labels to the parser
for _, annotations in TRAIN_DATA:
for dep in annotations.get('deps', []):
2017-10-26 16:15:37 +03:00
parser.add_label(dep)
# get names of other pipes to disable them during training
other_pipes = [pipe for pipe in nlp.pipe_names if pipe != 'parser']
2017-10-27 01:31:30 +03:00
with nlp.disable_pipes(*other_pipes): # only train parser
optimizer = nlp.begin_training()
2017-10-26 16:15:37 +03:00
for itn in range(n_iter):
random.shuffle(TRAIN_DATA)
losses = {}
for text, annotations in TRAIN_DATA:
nlp.update([text], [annotations], sgd=optimizer, losses=losses)
2017-10-26 16:15:37 +03:00
print(losses)
# test the trained model
test_text = "I like securities."
doc = nlp(test_text)
print('Dependencies', [(t.text, t.dep_, t.head.text) for t in doc])
# save model to output directory
if output_dir is not None:
output_dir = Path(output_dir)
if not output_dir.exists():
output_dir.mkdir()
nlp.to_disk(output_dir)
print("Saved model to", output_dir)
2017-10-27 04:55:04 +03:00
# test the saved model
2017-10-26 16:15:37 +03:00
print("Loading from", output_dir)
nlp2 = spacy.load(output_dir)
doc = nlp2(test_text)
print('Dependencies', [(t.text, t.dep_, t.head.text) for t in doc])
2016-10-16 18:05:55 +03:00
if __name__ == '__main__':
2017-10-26 16:15:37 +03:00
plac.call(main)
# expected result:
# [
# ('I', 'nsubj', 'like'),
# ('like', 'ROOT', 'like'),
# ('securities', 'dobj', 'like'),
# ('.', 'punct', 'like')
# ]