Basic examples (1)
Softanza
aX = [ [0,0], [0,1], [1,0], [1,1] ]
aY = [ [0], [1], [1], [0] ]
oN = new stzNeuralNetwork([ :Inputs = 2 ])
oN.AddDenseLayer(4, :Tanh)
oN.AddDenseLayer(1, :Sigmoid)
oT = new stzTrainer()
oT.SetLearningRate(0.5)
oT.Train(oN, aX, aY, 2000)
# 00 -> 0
? oN.Predict([0,0])[1] < 0.5
# 01 -> 1
? oN.Predict([0,1])[1] > 0.5
# 10 -> 1
? oN.Predict([1,0])[1] > 0.5
# 11 -> 0
? oN.Predict([1,1])[1] < 0.5
# the loss went down
? oT.FinalLoss() < oT.LossHistory()[1]
# ...essentially to zero
? oT.FinalLoss() < 0.001
# one loss per epoch was recorded
? len(oT.LossHistory())
# Why() still narrates
? StzFindFirst("trained 2000 epoch", oT.Why()) > 0Output
1 1 1 1 1 1 2000 1
run on 2026-10-02 04:23 inside the library at commit 0e72e2e2c; taken from numeric_backprop_narrated.ring(opens in a new tab)
The extensions of this method
Each extension, and whether it exists for this method; an extension that exists combines with the others (FindSTCS).
Extension
Exists
What it adds · The names
Q
yes
does what the method does, then returns the object, so the call can be chained · AddDenseLayerQ
QQ
no
chains on the next, more specific type of object (the Q ladder)
QQQ
no
chains on the most specific type of object (the Q ladder)
QRT
no
returns the result as an object of the requested type
CS
no
takes a case-sensitivity flag
ST
no
takes the position to start from (StartingAt)
IB
no
takes bounds that are included (IncludingBounds)
XT
no
the extended form: more parameters than the base method
XTT
no
yet another extension: a further extended form, after XT
Z
no
includes the position in what it returns
ZZ
no
returns positions as sections, [start, end]
W
no
selects by a condition
WF
no
selects by a condition, with the expressive keywords (@NextItem, @PreviousItem...)
D
no
takes a direction, forward or backward (Directional)
F
no
takes a function (the condition or the update is given as a function); the forms document also uses F for the future form, which defers the action
Many
no
works on a collection of items instead of one (the plural form)
Except
no
takes exceptions, what to leave out (the exceptional form)
U
no
returns the result without duplication
Every example is a test file of the library, run inside the library for this page; only those whose output kept every promise their file wrote are shown.
See also
- The class stzNeuralNetwork
- The area's reference: NLP & linguistics
- The other methods in these examples: stzNeuralNetwork.Predict · stzTrainer.FinalLoss · stzTrainer.LossHistory · stzTrainer.SetLearningRate · stzTrainer.Train · stzTrainer.Why

