DSIP-UPatras/ICASSP2019_TCN

hello, I have some problems about "run.sh"

Closed this issue · 3 comments

fornote commented 
       Every file could be run individually except **"attention_vis.py"** whose traceback was IndexError: list index out of range at line 40, **"RF = int(sys.argv[1])"**. 
       I have already downloaded the Ninapro DB1 data set. And I got the data by runing **"create_dataset.py"** before I ran the **"run.sh"**. However, when I ran the "run.sh", the Pycharm told me that **"ModuleNotFoundError: No module named 'tensorflow'"** in **File "script_experiment.py", line 3**. Because "script_experiment.py" could be run successfully and individually, I didn't know what had I done wrongly. Could you give me some help? thanks.
fornote commented

I have solved the second problem by running "run.sh" in Terminal instead of Pycharm.

ptsinganos commented

For the error in "attention_vis.py" you should run it with two arguments the receptive field (300 or 2500) and the subject id. For example, run:
python3 attention_vis.py 300 25
to visualize the attention weights of the subject 25 for receptive field = 300

fornote commented

For the error in "attention_vis.py" you should run it with two arguments the receptive field (300 or 2500) and the subject id. For example, run:
python3 attention_vis.py 300 25
to visualize the attention weights of the subject 25 for receptive field = 300

Thanks for your patiently reply. I have solved this problem, but another problem is bothering me.
After executing bash run.sh, 30 epoches of every subject were training finished, and the error, NameError: name 'MaskedConv1D' id not defined, occurred. Although there was the error, the coding could still continute to run. If I should loading a pre-training models?