Question about parameter adj_type?

[hustcxx]

I use commad as below to train model.
python train_abstractor.py --path ./cnn_data/model_path_concat/ --adj_type concat_triple --feat node_freq --bert --max_art 512 --max_abs 150 --batch 24 --docgraph --gpu_id 3

but get error like :
File "/data/home/chengxx/program/abstractive/AugumentSum_146/model/graph_enc.py", line 433, in forward
dim=0
RuntimeError: stack expects each tensor to be equal size, but got [74, 256] at entry 0 and [73, 256] at entry 1

Do I need to work with other parameters?