args = parser.parse_args()
args.cuda_device = [0] # [0, 1, 2, 3]
args.label_num = 2
args.threshold = 0.5
model_name = model_names[args.model]
torch.backends.cudnn.enabled = False
logger.debug('------------------------')
logger.debug('------------------------')
logger.debug('-------Parameters-------')
logger.debug('{}: {}'.format('model', model_name))
logger.debug(args)
# load data
# TODO: file path
reader = EventDataReader()
train_dataset = ensure_list(reader.read('./dataset/1_{}/{}/train.data'.format(args.file_dir, args.event_type)))
eval_dataset = ensure_list(reader.read('./dataset/1_{}/{}/eval.data'.format(args.file_dir, args.event_type)))
with open('./dataset/1_{}/{}/test.data'.format(args.file_dir, args.event_type), 'r', encoding='utf-8') as f:
test_dataset = json.loads(f.read())
# reader = EventDataReader()
# train_dataset = ensure_list(reader.read(os.path.join('.', 'dataset', 'example', 'train.data')))
# eval_dataset = ensure_list(reader.read(os.path.join('.', 'dataset', 'example', 'train.data')))
# with open(os.path.join('.', 'dataset', 'example', 'train.data'), 'r', encoding='utf-8') as f:
# test_dataset = json.loads(f.read())
# get vocabulary and embedding
vocab = Vocabulary.from_instances(train_dataset + eval_dataset,
min_count={"trigger_0": 0,
"trigger_agent_0": 0,
# init parameters
parser = argparse.ArgumentParser()
args = parser.parse_args()
args.embedding_size = 300
args.learning_rate = 1e-4
args.batch_size = 1
args.epochs = 10
args.patience = 1
args.cuda_device = -1
args.hidden_size = 8
args.hop_num = 6
args.label_num = 2
args.threshold = 0.5
# load data
reader = EventDataReader()
train_dataset = ensure_list(
reader.read(os.path.join('..', 'dataset', 'example', 'train.data')))
eval_dataset = ensure_list(
reader.read(os.path.join('..', 'dataset', 'example', 'train.data')))
test_dataset = ensure_list(
reader.read(os.path.join('..', 'dataset', 'example', 'train.data')))
# get vocabulary and embedding
vocab = Vocabulary.from_instances(train_dataset + eval_dataset,
min_count={
"trigger_0": 0,
"trigger_agent_0": 0,
"agent_attri_0": 0,
"trigger_object_0": 0,
"object_attri_0": 0,
event_type = [-1]
args.embedding_size = 300
args.learning_rate = 1e-3
args.cuda_device = [0]
args.hidden_size = 32
args.hop_num = 3
args.label_num = 2
args.threshold = 0.5
# 读取文件
for et in event_type:
reader = EventDataReader()
train_dataset = ensure_list(
reader.read(os.path.join('.', 'dataset', '1_4', str(et),
'train.data')))
eval_dataset = ensure_list(
reader.read(os.path.join('.', 'dataset', '1_4', str(et), 'eval.data')))
with open(os.path.join('.', 'dataset', '1_4', str(et), 'test.data'),
'r',
encoding='utf-8') as f:
test_dataset = json.loads(f.read())
print('load done')
vocab = Vocabulary.from_instances(train_dataset + eval_dataset,
min_count={
"trigger_0": 0,
"trigger_agent_0": 0,