You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I was considering to postprocess random search with the current data by default (in this case in the example experiment). Besides displaying a baseline, this has also the advantage to produce the standard output figures rather than the single-algorithm figures.
Using 'suite_name/.*randomsearch!' for this works on all suites but 'bbob-largescale' and 'sbox-cost', where no such data set exist.
The text was updated successfully, but these errors were encountered:
Thanks @nikohansen, I will produce these data sets then and put them into the data archive. I agree that every test suite should have this as a default baseline.
Will the random seed used appear with the baseline results?
To reproduce the exact results, we would probably also need the code (and possibly even the version number of the module creating the random numbers). Do you have a specific scenario in mind where this is relevant?
No scenario in particular, just an opinion.
If the results you produce are intended as baselines then they should be able to be reproduced exactly.
If it's too much trouble, I would much rather you move onto the next item on your collective todo lists. :)
I was considering to postprocess random search with the current data by default (in this case in the example experiment). Besides displaying a baseline, this has also the advantage to produce the standard output figures rather than the single-algorithm figures.
Using
'suite_name/.*randomsearch!'
for this works on all suites but'bbob-largescale'
and'sbox-cost'
, where no such data set exist.The text was updated successfully, but these errors were encountered: