if __name__ == "__main__":
config = Config()
processor = DataProcessor(config.vocab_size)
train_data, train_label = processor.get_datasets(config.train_pos_path,
config.train_neg_path,
config.vocab_path,
config.seq_len)
config.vocab_size = processor.vocab_size # 词表实际大小
test_data, test_label = processor.get_datasets(config.test_pos_path,
config.test_neg_path,
config.vocab_path,
config.seq_len)
train_set_iters = processor.batch_iter(list(zip(train_data, train_label)),
config.batch_size,
config.num_epochs,
shuffle=True)
test_batchs = processor.test_batchs(list(zip(test_data, test_label)),
config.batch_size)
model = TransformerEncoder(config)
model.to(config.device)
optimizer = torch.optim.Adam(model.parameters(), lr=config.lr)
loss_func = nn.BCELoss()
num_batches_per_epoch = int((len(train_data) - 1) / config.batch_size) + 1
model = train(model, train_set_iters, test_batchs, optimizer, loss_func,
config.device, num_batches_per_epoch)
