'NameError: global name 'batch_size_tensor' is not defined' in Paddle/benchmark/fluid/models/stacked_dynamic_lstm.py

[sneaxiy]

When the stacked_dynamic_lstm is running, the error "global name 'batch_size_tensor' is not defined" occurs.

$ python fluid_benchmark.py --model stacked_dynamic_lstm --device GPU --gpus 1
----------- Configuration Arguments -----------
batch_size: 32
cpus: 1
data_format: NCHW
data_path:
data_set: flowers
device: GPU
gpus: 1
infer_only: False
iterations: 80
learning_rate: 0.001
memory_optimize: False
model: stacked_dynamic_lstm
no_test: False
pass_num: 100
profile: False
skip_batch_num: 5
update_method: local
use_cprof: False
use_fake_data: False
use_nvprof: False
use_reader_op: False

Cache file /root/.cache/paddle/dataset/imdb/aclImdb_v1.tar.gz not found, downloading http://ai.stanford.edu/%7Eamaas/data/sentiment/aclImdb_v1.tar.gz
[==================================================]Traceback (most recent call last):
File "fluid_benchmark.py", line 452, in
main()
File "fluid_benchmark.py", line 412, in main
train_args = list(model_def.get_model(args))
File "/Paddle/Paddle/benchmark/fluid/models/stacked_dynamic_lstm.py", line 113, in get_model
target_vars=[batch_acc, batch_size_tensor])
NameError: global name 'batch_size_tensor' is not defined