spaCy/spacy/pipeline/senter.pyx
Daniël de Kok e5debc68e4
Tagger: use unnormalized probabilities for inference (#10197)
* Tagger: use unnormalized probabilities for inference

Using unnormalized softmax avoids use of the relatively expensive exp function,
which can significantly speed up non-transformer models (e.g. I got a speedup
of 27% on a German tagging + parsing pipeline).

* Add spacy.Tagger.v2 with configurable normalization

Normalization of probabilities is disabled by default to improve
performance.

* Update documentation, models, and tests to spacy.Tagger.v2

* Move Tagger.v1 to spacy-legacy

* docs/architectures: run prettier

* Unnormalized softmax is now a Softmax_v2 option

* Require thinc 8.0.14 and spacy-legacy 3.0.9
2022-03-15 14:15:31 +01:00

189 lines
6.4 KiB
Cython

# cython: infer_types=True, profile=True, binding=True
from typing import Optional, Callable
from itertools import islice
import srsly
from thinc.api import Model, SequenceCategoricalCrossentropy, Config
from ..tokens.doc cimport Doc
from .tagger import Tagger
from ..language import Language
from ..errors import Errors
from ..scorer import Scorer
from ..training import validate_examples, validate_get_examples
from ..util import registry
from .. import util
# See #9050
BACKWARD_OVERWRITE = False
default_model_config = """
[model]
@architectures = "spacy.Tagger.v2"
[model.tok2vec]
@architectures = "spacy.HashEmbedCNN.v2"
pretrained_vectors = null
width = 12
depth = 1
embed_size = 2000
window_size = 1
maxout_pieces = 2
subword_features = true
"""
DEFAULT_SENTER_MODEL = Config().from_str(default_model_config)["model"]
@Language.factory(
"senter",
assigns=["token.is_sent_start"],
default_config={"model": DEFAULT_SENTER_MODEL, "overwrite": False, "scorer": {"@scorers": "spacy.senter_scorer.v1"}},
default_score_weights={"sents_f": 1.0, "sents_p": 0.0, "sents_r": 0.0},
)
def make_senter(nlp: Language, name: str, model: Model, overwrite: bool, scorer: Optional[Callable]):
return SentenceRecognizer(nlp.vocab, model, name, overwrite=overwrite, scorer=scorer)
def senter_score(examples, **kwargs):
def has_sents(doc):
return doc.has_annotation("SENT_START")
results = Scorer.score_spans(examples, "sents", has_annotation=has_sents, **kwargs)
del results["sents_per_type"]
return results
@registry.scorers("spacy.senter_scorer.v1")
def make_senter_scorer():
return senter_score
class SentenceRecognizer(Tagger):
"""Pipeline component for sentence segmentation.
DOCS: https://spacy.io/api/sentencerecognizer
"""
def __init__(
self,
vocab,
model,
name="senter",
*,
overwrite=BACKWARD_OVERWRITE,
scorer=senter_score,
):
"""Initialize a sentence recognizer.
vocab (Vocab): The shared vocabulary.
model (thinc.api.Model): The Thinc Model powering the pipeline component.
name (str): The component instance name, used to add entries to the
losses during training.
scorer (Optional[Callable]): The scoring method. Defaults to
Scorer.score_spans for the attribute "sents".
DOCS: https://spacy.io/api/sentencerecognizer#init
"""
self.vocab = vocab
self.model = model
self.name = name
self._rehearsal_model = None
self.cfg = {"overwrite": overwrite}
self.scorer = scorer
@property
def labels(self):
"""RETURNS (Tuple[str]): The labels."""
# labels are numbered by index internally, so this matches GoldParse
# and Example where the sentence-initial tag is 1 and other positions
# are 0
return tuple(["I", "S"])
@property
def hide_labels(self):
return True
@property
def label_data(self):
return None
def set_annotations(self, docs, batch_tag_ids):
"""Modify a batch of documents, using pre-computed scores.
docs (Iterable[Doc]): The documents to modify.
batch_tag_ids: The IDs to set, produced by SentenceRecognizer.predict.
DOCS: https://spacy.io/api/sentencerecognizer#set_annotations
"""
if isinstance(docs, Doc):
docs = [docs]
cdef Doc doc
cdef bint overwrite = self.cfg["overwrite"]
for i, doc in enumerate(docs):
doc_tag_ids = batch_tag_ids[i]
if hasattr(doc_tag_ids, "get"):
doc_tag_ids = doc_tag_ids.get()
for j, tag_id in enumerate(doc_tag_ids):
if doc.c[j].sent_start == 0 or overwrite:
if tag_id == 1:
doc.c[j].sent_start = 1
else:
doc.c[j].sent_start = -1
def get_loss(self, examples, scores):
"""Find the loss and gradient of loss for the batch of documents and
their predicted scores.
examples (Iterable[Examples]): The batch of examples.
scores: Scores representing the model's predictions.
RETURNS (Tuple[float, float]): The loss and the gradient.
DOCS: https://spacy.io/api/sentencerecognizer#get_loss
"""
validate_examples(examples, "SentenceRecognizer.get_loss")
labels = self.labels
loss_func = SequenceCategoricalCrossentropy(names=labels, normalize=False)
truths = []
for eg in examples:
eg_truth = []
for x in eg.get_aligned("SENT_START"):
if x is None:
eg_truth.append(None)
elif x == 1:
eg_truth.append(labels[1])
else:
# anything other than 1: 0, -1, -1 as uint64
eg_truth.append(labels[0])
truths.append(eg_truth)
d_scores, loss = loss_func(scores, truths)
if self.model.ops.xp.isnan(loss):
raise ValueError(Errors.E910.format(name=self.name))
return float(loss), d_scores
def initialize(self, get_examples, *, nlp=None):
"""Initialize the pipe for training, using a representative set
of data examples.
get_examples (Callable[[], Iterable[Example]]): Function that
returns a representative sample of gold-standard Example objects.
nlp (Language): The current nlp object the component is part of.
DOCS: https://spacy.io/api/sentencerecognizer#initialize
"""
validate_get_examples(get_examples, "SentenceRecognizer.initialize")
util.check_lexeme_norms(self.vocab, "senter")
doc_sample = []
label_sample = []
assert self.labels, Errors.E924.format(name=self.name)
for example in islice(get_examples(), 10):
doc_sample.append(example.x)
gold_tags = example.get_aligned("SENT_START")
gold_array = [[1.0 if tag == gold_tag else 0.0 for tag in self.labels] for gold_tag in gold_tags]
label_sample.append(self.model.ops.asarray(gold_array, dtype="float32"))
assert len(doc_sample) > 0, Errors.E923.format(name=self.name)
assert len(label_sample) > 0, Errors.E923.format(name=self.name)
self.model.initialize(X=doc_sample, Y=label_sample)
def add_label(self, label, values=None):
raise NotImplementedError