[Rspamd-Users] [ext] Is my neural module working after updating from 1.9.4 to 2.0

[Tamás Németh]

Well, I assume, you are using Rspamd 1.9.4. I switched to 2.0 where
the neural module is heavily rewritten, and it doesn't produce such a
verbose logging as the previous version. I fact it doesn't even
produce any log messages when debugging of the neural module is not
enabled explicitly. Anyways, I remember those messages in your mail
from the previous version, and I was also worrying because of them,
but despite this my neural module successfully finished collecting
enough samples in a few days/weeks, and at that point it started to
tag the mails with the appropriate symbols. So, I don't exactly know
what your log messages mean, but they didn't mean any critical problem
according to my observations.

Sincerely,

Németh, Tamás
IT sysadmin
Some Company in Hungary

Ralf Hildebrandt <Ralf.Hildebrandt at charite.de> ezt írta (időpont:
2019. okt. 17., Cs, 18:19):
>
> I'm also under the impression it's not working as it should:
>
> Oct 17 14:27:29 mail-cbf rspamd[56117]: <ac8e09>; neural; neural.lua:337: LONG:default:14 ann score: nan
> Oct 17 14:27:29 mail-cbf rspamd[56117]: <ac8e09>; neural; neural.lua:337: SHORT:default:522 ann score: nan
> Oct 17 14:27:29 mail-cbf rspamd[56117]: <ac8e09>; lua; neural.lua:455: cannot learn ANN LONG:default: too many ham samples: 5501
> Oct 17 14:27:29 mail-cbf rspamd[56117]: <ac8e09>; lua; neural.lua:455: cannot learn ANN SHORT:default: too many ham samples: 111