def train(params):
global checkpoint_dir, ckpt, model
assert params["mode"].lower() == "train", "change training mode to 'train'"
vocab = Vocab(params["vocab_path"], params["vocab_size"])
print('true vocab is ', vocab)
print("Creating the batcher ...")
b = batcher(vocab, params) # 下次PGN详细讲，仅仅是将数据封装成tf特定的格式
print("Building the model ...")
if params["model"] == "SequenceToSequence":
model = SequenceToSequence(params)
# elif params["model"] == "PGN":
# model = PGN(params)
print("Creating the checkpoint manager")
if params["model"] == "SequenceToSequence":
checkpoint_dir = "{}/checkpoint".format(params["seq2seq_model_dir"])
ckpt = tf.train.Checkpoint(step=tf.Variable(0),
SequenceToSequence=model)
# elif params["model"] == "PGN":
# checkpoint_dir = "{}/checkpoint".format(params["pgn_model_dir"])
# ckpt = tf.train.Checkpoint(step=tf.Variable(0), PGN=model)
ckpt_manager = tf.train.CheckpointManager(ckpt,
checkpoint_dir,
max_to_keep=5)
ckpt.restore(ckpt_manager.latest_checkpoint)
if ckpt_manager.latest_checkpoint:
print("Restored from {}".format(ckpt_manager.latest_checkpoint))
else:
print("Initializing from scratch.")
print("Starting the training ...")
train_model(model, b, params, ckpt_manager)
def train(params):
assert params["mode"].lower() == "train", "change training mode to 'train'"
vocab = Vocab(params["vocab_path"], params["vocab_size"])
print('true vocab is ', vocab)
print("Creating the batcher ...")
b = batcher(vocab, params)
print("Building the model ...")
# model = SequenceToSequence(params)
# model = PGN(params)
model = PGN_TRANSFORMER(params)
print("Creating the checkpoint manager")
checkpoint_dir = "{}/checkpoint".format(params["seq2seq_model_dir"])
# 防止训练过程中中断，不需要重新训练
ckpt = tf.train.Checkpoint(SequenceToSequence=model)
ckpt_manager = tf.train.CheckpointManager(ckpt, checkpoint_dir, max_to_keep=5)
ckpt.restore(ckpt_manager.latest_checkpoint)
if ckpt_manager.latest_checkpoint:
print("Restored from {}".format(ckpt_manager.latest_checkpoint))
else:
print("Initializing from scratch.")
print("Starting the training ...")
train_model(model, b, params, ckpt, ckpt_manager)
def train_and_test(data_dir, no_of_epochs=4):
results_dir = data_dir + '/results'
batch_size = 64
vgg = Vgg16()
train_model(vgg, data_dir, batch_size, no_of_epochs)
batches, preds = test_model(vgg, data_dir + '/test', batch_size=batch_size)
save_array(results_dir + '/test_preds', preds)
save_array(results_dir + '/filenames', batches.filenames)
return batches, preds, vgg
import pandas
import torch
from model.DeepCoNN import DeepCoNNConfig, DeepCoNN
from utils.data_reader import get_train_dev_test_data
from utils.train_helper import train_model
from utils.word2vec_hepler import WORD_EMBEDDING_SIZE, load_embedding_weights
train_data, dev_data, test_data = get_train_dev_test_data()
know_data = pandas.concat([train_data, dev_data])
config = DeepCoNNConfig(
num_epochs=50,
batch_size=16,
learning_rate=1e-3,
l2_regularization=1e-2,
learning_rate_decay=0.95,
device="cuda:0" if torch.cuda.is_available() else "cpu",
max_review_length=
2048, # Make sure this value is smaller than max_length in data_reader.py
word_dim=WORD_EMBEDDING_SIZE,
kernel_widths=[2, 3, 5, 7],
kernel_deep=100,
latent_factors=50,
fm_k=8)
model = DeepCoNN(config, load_embedding_weights())
train_model(model, train_data, dev_data)