使用OpenKE中TransH模型在自制数据集训练时遇到错误

Question

各位大佬好，最近在用OpenKE中TransH模型在自制简单数据集上训练时遇到了麻烦，在用GPU时报“CUDA ERROR: device-side assert triggered”的错误，在用CPU时报“IndexError: index out of range in self”的错误，我猜测这应该是哪个维度出错了，但一直找不到原因，也不知道该改哪里，下面是我的训练代码和完整的运行情况，希望各位大佬多多指正：
训练代码如下：

import openke
from openke.config import Trainer, Tester
from openke.module.model import TransH
from openke.module.loss import MarginLoss
from openke.module.strategy import NegativeSampling
from openke.data import TrainDataLoader, TestDataLoader

# dataloader for training
train_dataloader = TrainDataLoader(
    in_path = "./benchmarks/XG02/",
    nbatches = 24,
    threads = 8, 
    sampling_mode = "normal", 
    bern_flag = 1, 
    filter_flag = 1, 
    neg_ent = 1,
    neg_rel = 0)

# dataloader for test
test_dataloader = TestDataLoader("./benchmarks/XG02/", "link")

# define the model
transh = TransH(
    ent_tot = train_dataloader.get_ent_tot(),
    rel_tot = train_dataloader.get_rel_tot(),
    dim = 24, 
    p_norm = 1, 
    norm_flag = True)

# define the loss function
model = NegativeSampling(
    model = transh, 
    loss = MarginLoss(margin = 4.0),
    batch_size = train_dataloader.get_batch_size()
)


# train the model
trainer = Trainer(model = model, data_loader = train_dataloader, train_times = 1000, alpha = 0.5, use_gpu = False)
print(model)
print(transh)
print(trainer.model.parameters)
trainer.run()
transh.save_checkpoint('./checkpoint/transh.ckpt')

# test the model
transh.load_checkpoint('./checkpoint/transh.ckpt')
tester = Tester(model = transh, data_loader = test_dataloader, use_gpu = False)
tester.run_link_prediction(type_constrain = False)

运行情况如下：

Answer 1

会不会是你的数据格式有问题，openke要求的(e1,e2,r)，而很多数据集确是(e1,r,e2)