RecursionError: maximum recursion depth exceeded

[niuqun]

Dear all,

I downloaded this project and ran it on my server, with Python 3.6.5 and PyTorch 0.4.1. My operating system is Ubuntu 16.04. When I execute the example command in example_command.sh (line 10 to line 13), it gives this error. Anyone else has encountered this before?

./train_pushed.py --num-train 100 --num-val 40 --num-test 100 --batch-size 64\ --network Inception --lr 1e-5 --feature-size 64 --reg 2e-3 --epochs 60 \ --results-freq 10 --val-freq 2 --num-batches 10\ --normalize-features --num-clusters 1 --loss Soft

Traceback (most recent call last):
File "./train_pushed.py", line 691, in
main()
File "./train_pushed.py", line 207, in main
model = Net(feature_size=feature_size, im_size=im_size, normalize=args.normalize_features)
File "/home/qniu/PycharmProjects/metric-learning-master/model_net.py", line 100, in init
init.xavier_normal(self.inception.fc.weight)
File "/home/qniu/PycharmProjects/metric-learning-master/init.py", line 220, in xavier_normal
xavier_normal(tensor.data, gain=gain)
File "/home/qniu/PycharmProjects/metric-learning-master/init.py", line 220, in xavier_normal
xavier_normal(tensor.data, gain=gain)
File "/home/qniu/PycharmProjects/metric-learning-master/init.py", line 220, in xavier_normal
xavier_normal(tensor.data, gain=gain)
[Previous line repeated 986 more times]
File "/home/qniu/PycharmProjects/metric-learning-master/init.py", line 219, in xavier_normal
if isinstance(tensor, Variable):
RecursionError: maximum recursion depth exceeded