Skip to first unread message
wan...@vt.edu
Dec 5, 2018, 10:55:26 AM12/5/18
to aenet
Hi,
Is anyone knows how to monitor the real-time training results? The RMSE will be written out to the "std.o" file if and only if the training is stopped.
Besides manually use "O.75t-75t.nn-02239" with prediction.x to get the training result, is any method to get the real-time RMSE?
Moreover, what format .nn file is? I tried to use binary viewer to check it, but it shows garbled text.
Thanks a lot. :)
Nongnuch Artrith
Dec 6, 2018, 1:29:34 PM12/6/18
to wan...@vt.edu, aenet
Hi Wang, Shih-Han:
'train.x' writes all output to standard out, which on your computer system might be a file named 'std.o'. You can redirect standard out according to your operation system. Depending on your Fortran compiler, the output might be buffered, i.e., the output file will only be updated after a certain amount of output has been accumulated. Your compiler's manual will tell you how to disable I/O buffering if you want to do that. For example, for the GNU Fortran compiler buffering can be disabled by setting the environment variable 'GFORTRAN_UNBUFFERED_ALL' to 'y'.
All the best,
Nong
P.S. In the future, we would appreciate if you would sign with your name and affiliation.
'train.x' writes all output to standard out, which on your computer system might be a file named 'std.o'. You can redirect standard out according to your operation system. Depending on your Fortran compiler, the output might be buffered, i.e., the output file will only be updated after a certain amount of output has been accumulated. Your compiler's manual will tell you how to disable I/O buffering if you want to do that. For example, for the GNU Fortran compiler buffering can be disabled by setting the environment variable 'GFORTRAN_UNBUFFERED_ALL' to 'y'.
All the best,
Nong
P.S. In the future, we would appreciate if you would sign with your name and affiliation.
--
Reply all
Reply to author
Forward
0 new messages