Rooba
/
spaCy
mirror of https://github.com/explosion/spaCy.git
1
1
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Fix training with preset vectors

This commit is contained in:
Matthew Honnibal 2017-09-22 20:00:40 -05:00
parent 05596159bf
commit e93d43a43a
1 changed files with 10 additions and 35 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

45
spacy/cli/train.py
View File

@ -30,14 +30,14 @@ from ..compat import json_dumps
n_iter=("number of iterations", "option", "n", int), n_iter=("number of iterations", "option", "n", int),
n_sents=("number of sentences", "option", "ns", int), n_sents=("number of sentences", "option", "ns", int),
use_gpu=("Use GPU", "option", "g", int), use_gpu=("Use GPU", "option", "g", int),
resume=("Whether to resume training", "flag", "R", bool), vectors=("Model to load vectors from", "option", "v"),
no_tagger=("Don't train tagger", "flag", "T", bool), no_tagger=("Don't train tagger", "flag", "T", bool),
no_parser=("Don't train parser", "flag", "P", bool), no_parser=("Don't train parser", "flag", "P", bool),
no_entities=("Don't train NER", "flag", "N", bool), no_entities=("Don't train NER", "flag", "N", bool),
gold_preproc=("Use gold preprocessing", "flag", "G", bool), gold_preproc=("Use gold preprocessing", "flag", "G", bool),
) )
def train(cmd, lang, output_dir, train_data, dev_data, n_iter=20, n_sents=0, def train(cmd, lang, output_dir, train_data, dev_data, n_iter=20, n_sents=0,
use_gpu=-1, resume=False, no_tagger=False, no_parser=False, no_entities=False, use_gpu=-1, vectors=None, no_tagger=False, no_parser=False, no_entities=False,
gold_preproc=False): gold_preproc=False):
""" """
Train a model. Expects data in spaCy's JSON format. Train a model. Expects data in spaCy's JSON format.
@ -73,25 +73,20 @@ def train(cmd, lang, output_dir, train_data, dev_data, n_iter=20, n_sents=0,
corpus = GoldCorpus(train_path, dev_path, limit=n_sents) corpus = GoldCorpus(train_path, dev_path, limit=n_sents)
n_train_words = corpus.count_train() n_train_words = corpus.count_train()
if not resume: lang_class = util.get_lang_class(lang)
lang_class = util.get_lang_class(lang) nlp = lang_class(pipeline=pipeline)
nlp = lang_class(pipeline=pipeline) if vectors:
optimizer = nlp.begin_training(lambda: corpus.train_tuples, device=use_gpu) util.load_model(vectors, vocab=nlp.vocab)
else: optimizer = nlp.begin_training(lambda: corpus.train_tuples, device=use_gpu)
print("Load resume")
util.use_gpu(use_gpu)
nlp = _resume_model(lang, pipeline, corpus)
optimizer = nlp.resume_training(device=use_gpu)
lang_class = nlp.__class__
nlp._optimizer = None nlp._optimizer = None
print("Itn.\tLoss\tUAS\tNER P.\tNER R.\tNER F.\tTag %\tToken %") print("Itn.\tLoss\tUAS\tNER P.\tNER R.\tNER F.\tTag %\tToken %")
try: try:
train_docs = corpus.train_docs(nlp, projectivize=True, noise_level=0.0,
gold_preproc=gold_preproc, max_length=0)
train_docs = list(train_docs)
for i in range(n_iter): for i in range(n_iter):
with tqdm.tqdm(total=n_train_words, leave=False) as pbar: with tqdm.tqdm(total=n_train_words, leave=False) as pbar:
train_docs = corpus.train_docs(nlp, projectivize=True, noise_level=0.0,
gold_preproc=gold_preproc, max_length=0)
losses = {} losses = {}
for batch in minibatch(train_docs, size=batch_sizes): for batch in minibatch(train_docs, size=batch_sizes):
docs, golds = zip(*batch) docs, golds = zip(*batch)
@ -124,26 +119,6 @@ def train(cmd, lang, output_dir, train_data, dev_data, n_iter=20, n_sents=0,
except: except:
pass pass
def _resume_model(lang, pipeline, corpus):
nlp = util.load_model(lang)
pipes = {getattr(pipe, 'name', None) for pipe in nlp.pipeline}
for name in pipeline:
if name not in pipes:
factory = nlp.Defaults.factories[name]
for pipe in factory(nlp):
if hasattr(pipe, 'begin_training'):
pipe.begin_training(corpus.train_tuples,
pipeline=nlp.pipeline)
nlp.pipeline.append(pipe)
nlp.meta['pipeline'] = pipeline
if nlp.vocab.vectors.data.shape[1] >= 1:
nlp.vocab.vectors.data = Model.ops.asarray(
nlp.vocab.vectors.data)
return nlp
def _render_parses(i, to_render): def _render_parses(i, to_render):
to_render[0].user_data['title'] = "Batch %d" % i to_render[0].user_data['title'] = "Batch %d" % i
with Path('/tmp/entities.html').open('w') as file_: with Path('/tmp/entities.html').open('w') as file_: