Closed alix-tz closed 1 year ago
Just a follow up, for now I'm using -s
(as mentioned in #397).
I've set the deterministic mode to 'warn' now. If you want truly deterministic training for recognition you'll have to skip CUDA as the CTC loss doesn't have a deterministic implementation but the differences between two runs are probably negligible.
On Kraken 4.3.7, using this command:
ketos -r train -o $something $gt -f binary {other options}
I get this:
As mentioned in #302 , it seems that using ketos's
-r
is preferable over using-s 134
alone.Do you have any recommandation to make reproducible trainings? And incidentally, any recommandation on how to make my command work? Is there anything I can do about it?