How to tell where my code gives an exp overflow, in numpy?

Question

I keep getting this error when using numpy:

Warning (from warnings module):
  File "somepath.py", line 249
    return 1 / ( 1 + np.exp( -x))
RuntimeWarning: overflow encountered in exp

However this doesn't tell me anything, as I still have no idea how it got to this error. Usually when I get an error it traces back through the functions it went through to get to the error, but not with exp overflow.

How do I get an error in which it does show me this?

please refer http://stackoverflow.com/questions/9559346/deal-with-overflow-in-exp-using-numpy — JkShaw, Apr 26 '17 at 14:29

score 2 · Accepted Answer · answered Apr 26 '17 at 14:36

2

You can ask Python to turn the warning into an error:

import warnings    
warnings.filterwarnings("error", category=RuntimeWarning)

(Docs)

answered Apr 26 '17 at 14:36

Stephen Rauch

47,830
31
106
135

Brilliant. Thanks – Perm. Questiin Apr 26 '17 at 14:59

How to tell where my code gives an exp overflow, in numpy?

1 Answers1