Train result differently when run from train_bash.sh with running directly form the train.py

[joyicejin]

when run the train.py directly
the train result in fold0 (for example) directory tree is --fold0

--deepfeaturenet --train_summary (fold) --perf_fold0.npz --checkpoint --params_fold0.npz --perf_fold0.npz .........and so on

--deepsleepnet --train_summary (fold) --perf_fold0.npz --checkpoint --params_fold0.npz --perf_fold0.npz .........and so on

when use the argument:./batch_train.sh data/eeg_fpz_cz/ output 20 0 19 0 the train result in fold0 directory tree is --fold0 --deepfeaturenet (only has one fold) --train_summary --perf_fold0.npz

The difference is when the bash files result lack deepsleepnet fold and perf_fold0.npz I just wonder why the result is different when use bash files. Could you help me?

[akaraspt]

Can you tell me more about the command that you use to run train.py directly?

[joyicejin]

My os is win10, I only changed the path in the train.py, use the pycharm to run, and using the default settings to train the fold. but I use the command batch_train.sh data\eeg_fpz_cz\ output 20 0 19 0, the error happens

[joyicejin]

Thanks for your comments, because of the bad command, solve it and closed