Ejemplos de JointParser.train en Python

Ejemplo n.º 1

Archivo: double_cnnrefine.py Proyecto: xlzwhboy/bert-2019

# -*- coding:utf-8 -*-
# Author: hankcs
# Date: 2019-01-18 23:44
from bertsota.common.utils import init_logger
from bertsota.parser.biaffine_parser import RefineParser, CNNParser
from bertsota.parser.joint_parser import JointParser

parser = JointParser(cls_parser=CNNParser)
save_dir = 'data/model/2cnn-refine-joint-sdp'
parser.train(
    train_file=[
        'data/semeval15/en.dm.train.conllu',
        'data/semeval15/en.pas.train.conllu',
        'data/semeval15/en.psd.train.conllu'
    ],
    dev_file=[
        'data/semeval15/en.dm.dev.conllu', 'data/semeval15/en.pas.dev.conllu',
        'data/semeval15/en.psd.dev.conllu'
    ],
    save_dir=save_dir,
    pretrained_embeddings_file='data/embedding/glove.6B.100d.shrinked.txt',
    lstm_hiddens=800,
    root='root')
parser.load(save_dir)
logger = init_logger(save_dir, 'test.log')
parser.evaluate(test_files=[
    'data/semeval15/en.id.dm.conllu', 'data/semeval15/en.id.pas.conllu',
    'data/semeval15/en.id.psd.conllu'
],
                save_dir=save_dir,
                logger=logger)
parser.evaluate(test_files=[

Ejemplo n.º 2

from bertsota.common.utils import init_logger
from bertsota.parser.biaffine_parser import RefineParser, SharedRNNParser
from bertsota.parser.joint_parser import JointParser

for ratio in range(40, 101, 20):
    parser = JointParser(cls_parser=SharedRNNParser)
    save_dir = 'data/model/transfer-pas-psd-dm-{}'.format(ratio)
    print(save_dir)
    parser.train(
        train_file=[
            'data/semeval15/en.pas.train.{}.conllu'.format(ratio),
            'data/semeval15/en.psd.train.{}.conllu'.format(ratio),
            'data/semeval15/en.dm.train.{}.conllu'.format(ratio)
        ],
        dev_file=[
            'data/semeval15/en.pas.dev.conllu',
            'data/semeval15/en.psd.dev.conllu',
            'data/semeval15/en.dm.dev.conllu'
        ],
        save_dir=save_dir,
        pretrained_embeddings_file='data/embedding/glove.6B.100d.shrinked.txt',
        lstm_hiddens=800,
        transfer='data/model/pas-psd-sharernn',
        root='root')
    logger = init_logger(save_dir, 'test.log')
    parser.evaluate(test_files=[
        'data/semeval15/en.id.pas.conllu', 'data/semeval15/en.id.psd.conllu',
        'data/semeval15/en.id.dm.conllu'
    ],
                    save_dir=save_dir,
                    logger=logger)
    parser.evaluate(test_files=[

Ejemplo n.º 3

# Date: 2019-01-16 22:03
from bertsota.common.utils import init_logger
from bertsota.parser.biaffine_parser import RefineParser
from bertsota.parser.joint_parser import JointParser

parser = JointParser(cls_parser=RefineParser)
save_dir = 'data/model/refine-bert-large'
parser.train(
    train_file=[
        'data/semeval15/en.dm.train.conllu',
        'data/semeval15/en.pas.train.conllu',
        'data/semeval15/en.psd.train.conllu'
    ],
    dev_file=[
        'data/semeval15/en.dm.dev.conllu', 'data/semeval15/en.pas.dev.conllu',
        'data/semeval15/en.psd.dev.conllu'
    ],
    save_dir=save_dir,
    lstm_hiddens=1200,
    pretrained_embeddings_file='data/embedding/glove.6B.100d.shrinked.txt',
    bert_path=[
        'data/embedding/en.train.large.uncased.bert',
        'data/embedding/en.dev.large.uncased.bert'
    ],
    root='root')
parser.load(save_dir)
logger = init_logger(save_dir, 'test.log')
parser.evaluate(test_files=[
    'data/semeval15/en.id.dm.conllu', 'data/semeval15/en.id.pas.conllu',
    'data/semeval15/en.id.psd.conllu'
],
                bert_path='data/embedding/en.id.large.uncased.bert',