[Rspamd-Users] Neural Network manually training

SysAdmin EM emawata at gmail.com
Wed Sep 22 17:38:40 UTC 2021


Hello, I am trying to manually train the neural network since I see that
many emails that are SPAM are being cataloged as NEURAL_HAM.

This is my config:

servers = "127.0.0.1:6379";

train {
  max_trains = 1k; # Number ham/spam samples needed to start train
  max_usages = 20; # Number of learn iterations while ANN data is valid
  learning_rate = 0.01; # Rate of learning
  max_iterations = 25; # Maximum iterations of learning (better preciseness
but also lower speed of learning)
}

ann_expire = 2d; # For how long ANN should be preserved in Redis

The documentation (https://rspamd.com/doc/modules/neural.html) indicates
that the files should be placed in /plugins/neural/learn. What would be the
exact path? in /etc/rspamd/local.d? or other path?

once the neural network has been trained, can automatic training be
deactivated? or do you always have to train?

Regards,


More information about the Users mailing list