Why didn't I save the model when I trained with the following command?Thank you very much and look forward to your reply

CUDA_VISIBLE_DEVICES='0' python -u train.py --network r100 --data-dir /home/xl/1mxnetinsight/insightface-master/datasets/20_faces_datasets --prefix ../save/ --ckpt 2 --target lfw --version-se 0

when I train,
...
call reset()
INFO:root:Epoch[11] Train-acc=0.166667
INFO:root:Epoch[11] Time cost=0.271
call reset()
INFO:root:Epoch[12] Train-acc=0.166667
INFO:root:Epoch[12] Time cost=0.258
call reset()
INFO:root:Epoch[13] Train-acc=0.500000
INFO:root:Epoch[13] Time cost=0.263

Start with train.py and reproduce the command using the supplied --prefix and --ckpt options, then inspect the training output and generated files after the shown epochs. Determine whether a checkpoint is written and what prevents the expected model from appearing; done means documenting a reproducible cause or the missing details needed to diagnose it.

Written by the indexing model from the issue text.