parser.add_argument('--seed', dest='seed', default=666, type=int,
help='随机种子') #
parser.add_argument('--gpu',
dest='gpu',
default=True,
type=bool,
help='是否使用gpu') #
parser.add_argument('--max_epoch',
dest='max_epoch',
default=20,
type=int,
help='最大训练epoch')
args = parser.parse_args() # 程序运行参数
config = Config() # 模型配置PYTORCH_TRANSFORMERS_CACHE
torch.manual_seed(args.seed)
if torch.cuda.is_available():
torch.cuda.manual_seed(args.seed)
def main():
trainset, validset, testset = [], [], []
if args.inference: # 测试时只载入测试集
with open(args.testset_path, 'r', encoding='utf8') as fr:
for line in fr:
testset.append(json.loads(line))
print(f'载入测试集{len(testset)}条')
else: # 训练时载入训练集和验证集
with open(args.trainset_path, 'r', encoding='utf8') as fr:
parser.add_argument('--seed', dest='seed', default=666, type=int,
help='随机种子') #
parser.add_argument('--gpu',
dest='gpu',
default=True,
type=bool,
help='是否使用gpu') #
parser.add_argument('--max_epoch',
dest='max_epoch',
default=40,
type=int,
help='最大训练epoch')
args = parser.parse_args() # 程序运行参数
config = Config() # 模型配置
torch.manual_seed(args.seed)
if torch.cuda.is_available():
torch.cuda.manual_seed(args.seed)
def main():
# 载入数据集
trainset, validset, testset = [], [], []
if args.inference: # 测试时只载入测试集
with open(args.testset_path, 'r', encoding='utf8') as fr:
for line in fr:
testset.append(json.loads(line))
print('载入测试集%d条' % len(testset))
dest='embed_path',
default='data/embed.txt',
type=str,
help='词向量位置')
parser.add_argument('--vad_path',
dest='vad_path',
default='data/vad.txt',
type=str,
help='vad位置')
parser.add_argument('--gpu',
dest='gpu',
default=True,
type=bool,
help='是否使用gpu')
args = parser.parse_args()
config = Config()
def filter_by_emotion(args):
vocab, embeds = [], []
with open(args.embed_path, 'r', encoding='utf8') as fr:
for line in fr:
line = line.strip()
word = line[:line.find(' ')]
vec = line[line.find(' ') + 1:].split()
embed = [float(v) for v in vec]
assert len(embed) == config.embedding_size # 检测词向量维度
vocab.append(word)
embeds.append(embed)
print(f'载入词汇表: {len(vocab)}个')
print(f'词向量维度: {config.embedding_size}')