max_usages = 20; # Number of learn iterations while ANN data is valid
spam_score = 8; # Score to learn spam
ham_score = -2; # Score to learn ham
+ learning_rate = 0.01; # Rate of learning (Torch only)
+ max_iterations = 25; # Maximum iterations of learning (Torch only)
}
timeout = 20; # Increase redis timeout
autotrain = true,
train_prob = 1.0,
learn_threads = 1,
+ learning_rate = 0.01,
},
use_settings = false,
per_user = false,
lim = lim + lim * 0.1
local exists = redis.call('SISMEMBER', KEYS[1], KEYS[2])
- if not exists or exists == 0 then
+ if not exists or tonumber(exists) == 0 then
redis.call('SADD', KEYS[1], KEYS[2])
end
local criterion = nn.MSECriterion()
local trainer = nn.StochasticGradient(anns[elt].ann_train,
criterion)
- trainer.learning_rate = 0.01
+ trainer.learning_rate = rule.train.learning_rate
trainer.verbose = false
trainer.maxIteration = rule.train.max_iterations
trainer.hookIteration = function(self, iteration, currentError)