Professional Version Basis of AI Backprop Hypertext Documentation
Copyright (c) 1990-97 by Donald R. Tveter
A Summary of Typed Commands
a <number> sets the momentum parameter, alpha
a <options> the algorithm command
ac <weight;> add connection
ah <layer> add hidden unit
b <options> benchmarking
c clear the network
cd <directory> changes the working directory to <directory>
ci <real> clear and initialize the network
ct <filename> makes a copy to <filename>
d <options> set delta-bar-delta parameters
e <number(s)> set the learning rate eta
f <options> lots of formatting options
i <filename> read input from the file
k <reals> kick the network
l <layer> list the values of units on that layer
m <numbers> make a network
onu <unit> turn on a unit
ofu <unit> turn off a unit
onw <weight> turn on a weight
ofw <weight> turn off a weight
p <options> list information about training patterns
pr <options> the predict command
pw <real> prune weights
q quit
qp <options> set quickprop parameters
r <options> run the training algorithm
rp <options> rprop parameters
rt <options> read the training set patterns
rw <filename> read the weights
rx <filename> read the extra training set patterns
s <seed> seed value
se <filename> save almost everything
si <option> increment seed
ss <options> supersab command
sw <filename> save weights
swe <int> save weights every so often
swem <option> save weights every test set minimum
t <options> list testing file statistics of various sorts
t <real> tolerance per output unit, between 0 and 1
tf <filename> gives the file name with testing patterns
to <real> tolerance overall
tpu <real> tolerance per output unit, any value
tr <int> special test for a recurrent network
trp <int> special test for a recurrent network
u <option> recurrent classification training set data
v <option> recurrent classification test set data
w <layer> <unit> list weights leading into unit