explosion/spaCy
1
1
Fork 0
mirror of https://github.com/explosion/spaCy.git synced 2025-01-23 15:54:13 +03:00
Code Issues Packages Projects Releases Wiki Activity

Update textcat example

Browse Source
This commit is contained in:
Matthew Honnibal 2017-10-04 15:12:28 +02:00
parent 39798b0172
commit f1b86dff8c
1 changed files with 12 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

17
examples/training/train_textcat.py
View File

@ -16,6 +16,11 @@ from spacy.gold import GoldParse, minibatch
from spacy.util import compounding from spacy.util import compounding
from spacy.pipeline import TextCategorizer from spacy.pipeline import TextCategorizer
# TODO: Remove this once we're not supporting models trained with thinc <6.9.0
import thinc.neural._classes.layernorm
thinc.neural._classes.layernorm.set_compat_six_eight(False)
def train_textcat(tokenizer, textcat, def train_textcat(tokenizer, textcat,
train_texts, train_cats, dev_texts, dev_cats, train_texts, train_cats, dev_texts, dev_cats,
@ -28,12 +33,13 @@ def train_textcat(tokenizer, textcat,
train_docs = [tokenizer(text) for text in train_texts] train_docs = [tokenizer(text) for text in train_texts]
train_gold = [GoldParse(doc, cats=cats) for doc, cats in train_gold = [GoldParse(doc, cats=cats) for doc, cats in
zip(train_docs, train_cats)] zip(train_docs, train_cats)]
train_data = zip(train_docs, train_gold) train_data = list(zip(train_docs, train_gold))
batch_sizes = compounding(4., 128., 1.001) batch_sizes = compounding(4., 128., 1.001)
for i in range(n_iter): for i in range(n_iter):
losses = {} losses = {}
train_data = tqdm.tqdm(train_data, leave=False) # Progress bar # Progress bar and minibatching
for batch in minibatch(train_data, size=batch_sizes): batches = minibatch(tqdm.tqdm(train_data, leave=False), size=batch_sizes)
for batch in batches:
docs, golds = zip(*batch) docs, golds = zip(*batch)
textcat.update(docs, golds, sgd=optimizer, drop=0.2, textcat.update(docs, golds, sgd=optimizer, drop=0.2,
losses=losses) losses=losses)
@ -65,12 +71,13 @@ def evaluate(tokenizer, textcat, texts, cats):
return {'textcat_p': precis, 'textcat_r': recall, 'textcat_f': fscore} return {'textcat_p': precis, 'textcat_r': recall, 'textcat_f': fscore}
def load_data(): def load_data(limit=0):
# Partition off part of the train data --- avoid running experiments # Partition off part of the train data --- avoid running experiments
# against test. # against test.
train_data, _ = thinc.extra.datasets.imdb() train_data, _ = thinc.extra.datasets.imdb()
random.shuffle(train_data) random.shuffle(train_data)
train_data = train_data[-limit:]
texts, labels = zip(*train_data) texts, labels = zip(*train_data)
cats = [(['POSITIVE'] if y else []) for y in labels] cats = [(['POSITIVE'] if y else []) for y in labels]
@ -90,7 +97,7 @@ def main(model_loc=None):
textcat = TextCategorizer(tokenizer.vocab, labels=['POSITIVE']) textcat = TextCategorizer(tokenizer.vocab, labels=['POSITIVE'])
print("Load IMDB data") print("Load IMDB data")
(train_texts, train_cats), (dev_texts, dev_cats) = load_data() (train_texts, train_cats), (dev_texts, dev_cats) = load_data(limit=1000)
print("Itn.\tLoss\tP\tR\tF") print("Itn.\tLoss\tP\tR\tF")
progress = '{i:d} {loss:.3f} {textcat_p:.3f} {textcat_r:.3f} {textcat_f:.3f}' progress = '{i:d} {loss:.3f} {textcat_p:.3f} {textcat_r:.3f} {textcat_f:.3f}'