all_time_start = datetime.now()
config = Config()
corpus = Processing(config.train_file, config.pre_trained_embed_file)
# train_sentences = Processing.data_handle(config.train_file)
dev_sentences = Processing.data_handle(config.dev_file, True)
test_sentences = Processing.data_handle(config.test_file, True)
print("\n训练预料：")
print("句子数：%d" % len(corpus.sentences))
print("词数：%d" % len(corpus.words))
print("字符数：%d" % len(corpus.chars))
print("词性数：%d" % len(corpus.tags))
lstm = LSTMTagger(corpus.word2id, corpus.char2id, corpus.tag2id,
corpus.embedding_matrix, config.embed_dim,
config.char_embed_dim, config.n_hidden)
train_data_loader = lstm.get_loader(dataset=corpus.load(config.train_file),
batch_size=config.batch_size,
thread_num=config.thread_num,
shuffle=config.shuffle)
dev_data_loader = lstm.get_loader(dataset=corpus.load(config.dev_file),
batch_size=config.batch_size,
thread_num=config.thread_num,
shuffle=config.shuffle)
test_data_loader = lstm.get_loader(dataset=corpus.load(config.test_file),
batch_size=config.batch_size,
thread_num=config.thread_num,
shuffle=config.shuffle)
optimizer = torch.optim.Adam(lstm.parameters(), lr=config.learn_rate)
loss_func = torch.nn.CrossEntropyLoss()
torch.set_num_threads(config.thread_num)
