def main():
# 读配置文件
config = parse_config()
# 载入训练集合
train_data = DataBatchIterator(config=config,
is_train=True,
dataset="train",
batch_size=config.batch_size,
shuffle=True)
train_data.load()
vocab = train_data.vocab
# 载入测试集合
test_data = DataBatchIterator(config=config,
is_train=False,
dataset="test",
batch_size=config.batch_size)
test_data.set_vocab(vocab)
test_data.load()
# 测试时
checkpoint = torch.load(config.save_model + ".pt",
map_location=config.device)
model = checkpoint
# model = build_textcnn_model(
# vocab, config, train=True)
predict, label = test_textcnn_model(model, test_data, config)
print(classification_report(label, predict))
def main():
# 读配置文件
config = parse_config()
# 载入训练集合
train_data = DataBatchIterator(
config=config,
is_train=True,
dataset="train",
#batch_size=config.batch_size,
shuffle=True)
train_data.load()
vocab = train_data.vocab
# 载入测试集合
valid_data = DataBatchIterator(
config=config,
is_train=False,
dataset="dev",
#batch_size=config.batch_size
)
valid_data.set_vocab(vocab)
valid_data.load()
# 构建textcnn模型
model = build_textcnn_model(vocab, config, train=True)
print(model)
# Do training.
padding_idx = vocab.stoi[PAD]
train_textcnn_model(model, train_data, valid_data, padding_idx, config)
torch.save(model, '%s.pt' % (config.save_model))
def main():
config = parse_config()
checkpoint = torch.load(config.save_model + ".pt",
map_location=config.device)
train_data = DataBatchIterator(config=config,
is_train=True,
dataset="train",
batch_size=config.batch_size,
shuffle=True)
train_data.load()
vocab = train_data.vocab
# 载入测试集合
valid_data = DataBatchIterator(config=config,
is_train=False,
dataset="test",
batch_size=config.batch_size)
valid_data.set_vocab(vocab)
valid_data.load()
# Do training.
padding_idx = vocab.stoi[PAD]
train_textcnn_model(checkpoint, train_data, valid_data, padding_idx,
config)
def main():
# 读配置文件
config = parse_config()
# 载入训练集合
train_data = DataBatchIterator(config=config,
is_train=True,
dataset="train",
batch_size=config.batch_size,
shuffle=True)
train_data.load()
vocab = train_data.vocab #词汇映射表
# 载入测试集合
test_data = DataBatchIterator(config=config,
is_train=False,
dataset="test",
batch_size=config.batch_size)
test_data.set_vocab(vocab)
test_data.load()
# 测试时载入模型
model = torch.load(config.save_model + ".pt", map_location=config.device)
print(model)
test(model, test_data)
def main():
# 读配置文件
config = parse_config()
# 载入训练集合
train_data = DataBatchIterator(
config=config,
is_train=True,
dataset="train",
batch_size=config.batch_size,
shuffle=True)
train_data.load()
vocab = train_data.vocab
# 载入测试集合
valid_data = DataBatchIterator(
config=config,
is_train=False,
dataset="dev",
batch_size=config.batch_size)
valid_data.set_vocab(vocab)
valid_data.load()
# 构建textcnn模型
model = build_textcnn_model(
vocab, config, train=True)
print(model)
# Do training.
padding_idx = vocab.stoi[PAD]
#train_textcnn_model(model, train_data,
# valid_data, padding_idx, config)
#torch.save(model, '%s.pt' % (config.save_model))
# 测试时
#加载测试集
test_data = DataBatchIterator(
config=config,
is_train=False,
dataset="test",
batch_size=config.batch_size)
test_data.load()
#读取训练好的模型
checkpoint = torch.load(config.save_model+".pt",
map_location = config.device)
#测试并打印评价
test_valid(checkpoint , config , test_data)
def main():
# 读配置文件
config = parse_config()
# 载入训练集合
train_data = DataBatchIterator(config=config,
is_train=True,
dataset="train",
batch_size=config.batch_size,
shuffle=True)
train_data.load()
vocab = train_data.vocab
# 载入测试集合
valid_data = DataBatchIterator(config=config,
is_train=False,
dataset="test",
batch_size=config.batch_size)
valid_data.set_vocab(vocab)
valid_data.load()
# 测试时
checkpoint = torch.load(config.save_model + ".pt",
map_location=config.device)
model = checkpoint
# model = build_textcnn_model(
# vocab, config, train=True)
model.eval()
total_loss = 0
valid_data_iter = iter(valid_data)
total_predict = torch.LongTensor([])
total_label = torch.LongTensor([])
for idx, batch in enumerate(valid_data_iter):
model.zero_grad()
ground_truth = batch.label
# batch_first = False
outputs = model(batch.sent)
predict = torch.max(outputs, 1)[1]
total_predict = torch.cat((total_predict, predict), 0)
total_label = torch.cat((total_label, batch.label), 0)
print(classification_report(total_label, total_predict))