def batch_predict_loader(config):
vocab = load_vocab(config.vocab)
label_dic = load_vocab(config.label_file)
test_data = read_corpus(config.test_file,
max_length=config.max_length,
label_dic=label_dic,
vocab=vocab)
test_ids = torch.LongTensor([temp.input_id
for temp in test_data]).to(config.device)
test_masks = torch.LongTensor([temp.input_mask
for temp in test_data]).to(config.device)
test_tags = torch.LongTensor([temp.label_id
for temp in test_data]).to(config.device)
batch_predict_dataset = TensorDataset(test_ids, test_masks)
batch_predict_load = DataLoader(batch_predict_dataset,
shuffle=True,
batch_size=config.batch_size)
test_dataset = TensorDataset(test_ids, test_masks, test_tags)
test_loader = DataLoader(test_dataset,
shuffle=True,
batch_size=config.batch_size)
return batch_predict_load, test_loader
def data_load(config):
vocab = load_vocab(config.vocab)
label_dic = load_vocab(config.label_file)
train_data = read_corpus(config.train_file,
max_length=config.max_length,
label_dic=label_dic,
vocab=vocab)
dev_data = read_corpus(config.dev_file,
max_length=config.max_length,
label_dic=label_dic,
vocab=vocab)
test_data = read_corpus(config.test_file,
max_length=config.max_length,
label_dic=label_dic,
vocab=vocab)
train_ids = torch.LongTensor([temp.input_id
for temp in train_data]).to(config.device)
train_masks = torch.LongTensor([temp.input_mask
for temp in train_data]).to(config.device)
train_tags = torch.LongTensor([temp.label_id
for temp in train_data]).to(config.device)
train_dataset = TensorDataset(train_ids, train_masks, train_tags)
train_loader = DataLoader(train_dataset,
shuffle=True,
batch_size=config.batch_size)
dev_ids = torch.LongTensor([temp.input_id
for temp in dev_data]).to(config.device)
dev_masks = torch.LongTensor([temp.input_mask
for temp in dev_data]).to(config.device)
dev_tags = torch.LongTensor([temp.label_id
for temp in dev_data]).to(config.device)
dev_dataset = TensorDataset(dev_ids, dev_masks, dev_tags)
dev_loader = DataLoader(dev_dataset,
shuffle=True,
batch_size=config.batch_size)
test_ids = torch.LongTensor([temp.input_id
for temp in test_data]).to(config.device)
test_masks = torch.LongTensor([temp.input_mask
for temp in test_data]).to(config.device)
test_tags = torch.LongTensor([temp.label_id
for temp in test_data]).to(config.device)
test_dataset = TensorDataset(test_ids, test_masks, test_tags)
test_loader = DataLoader(test_dataset,
shuffle=True,
batch_size=config.batch_size)
return train_loader, dev_loader, test_loader
def __init__(self, config, embedding_dim, hidden_dim, rnn_layers,
dropout_ratio, dropout1):
super(BERT_LSTM_CRF, self).__init__()
self.device = config.device
self.num_tags = len(load_vocab(config.label_file))
self.embedding_dim = embedding_dim
self.hidden_dim = hidden_dim
self.encoder = BertEmbedder.create()
self.lstm = nn.LSTM(self.embedding_dim,
self.hidden_dim,
num_layers=rnn_layers,
bidirectional=True,
dropout=dropout_ratio,
batch_first=True)
self.rnn_layers = rnn_layers
self.dropout1 = nn.Dropout(p=dropout1)
self.linear = nn.Linear(hidden_dim * 2, self.num_tags)
self.crf = CRF(num_tags=self.num_tags, device=self.device)
self.init_weights()
def __init__(self, config, embed_dim, dropout1):
super(BERT_CRF, self).__init__()
self.device = config.device
self.num_tags = len(load_vocab(config.label_file))
self.embed_dim = embed_dim
self.encoder = BertEmbedder.create()
self.dropout1 = nn.Dropout(p=dropout1)
self.linear = nn.Linear(self.embed_dim, self.num_tags)
self.crf = CRF(num_tags=self.num_tags, device=self.device)
def __init__(self, config, embedding_dim, dropout1):
super(BERT_ATTN_CRF, self).__init__()
self.device = config.device
self.num_tags = len(load_vocab(config.label_file))
self.encoder = BertEmbedder.create()
self.attention = Attn(config)
self.dropout = nn.Dropout(dropout1)
config.update(attention_dim=embedding_dim)
self.linear = nn.Linear(embedding_dim, self.num_tags)
self.crf = CRF(num_tags=self.num_tags, device=self.device)