[Rspamd-Users] rspamd - autolearn - check_balance = true

Andreas Wass - Glas Gasperlmair a.wass at glas-gasperlmair.at
Fri Oct 29 05:34:29 UTC 2021

Ich versuchs mal so:

Keiner Erfahrungen mit autolearn und der Balance zwischen gerlernten 

vg, Andi

Am 28.10.2021 um 08:14 schrieb Andreas Wass - Glas Gasperlmair:
> would say, i use rspamd with autolearn for a week now (not 
> check_balance = true)
> Andi
> Am 28.10.2021 um 07:37 schrieb Andreas Wass - Glas Gasperlmair:
>> I use it for a week now and as you can see below it is out of balance
>> BAYES_SPAM    redis    199
>> BAYES_HAM    redis    4170
>> Is it really recommended, that also Bayes-Ham- and Spam should be in 
>> balance not only ham/spam in neural module?
>> If yes, how can i reset this, to start learning manually from scratch?
>> best regards Andi
>> Am 27.10.2021 um 18:10 schrieb Andy Balholm:
>>> No, I'm not using it. I hadn't heard of it, because it isn't 
>>> mentioned in the Autolearning section of the Statistics module 
>>> documentation (just in a sample configuration farther up). It sounds 
>>> like a good idea, but I wouldn't recommend using it if your spam and 
>>> ham learns are already seriously out of balance; in that case, it 
>>> would basically disable autolearn for whichever you have way more of.

More information about the Users mailing list