diff options
Diffstat (limited to 'doc/ml.1')
| -rw-r--r-- | doc/ml.1 | 47 |
1 files changed, 41 insertions, 6 deletions
@@ -9,7 +9,8 @@ ml \- manual page for ml 0.1 .B ml \fI\,predict \/\fR[\fI\,-o FILE\/\fR] \fI\,FILE\/\fR .SH DESCRIPTION -Train and predict json data +ml is a simple neural network maker made to train and predict JSON data, +it is suitable to work on classification problems. .SH OPTIONS .TP \fB\-h\fR, \fB\-\-help\fR @@ -23,13 +24,47 @@ Epochs to train the model (only works with train) .TP \fB\-o\fR, \fB\-\-output\fR=\fI\,FILE\/\fR Output file (only works with predict) +.TP +\fB\-c\fR, \fB\-\-config\fR=\fI\,FILE\/\fR +Configuration filepath [default=~/.config/ml/ml.cfg] +.SH FILES +~/.config/ml/ml.cfg + File path for network configuration, here you can setup the network + architecture and its training parameters .SH EXAMPLES -.IP -\f(CW$ ml train -e 150 -a 1e-4 housing.json\fR -.br -\f(CW$ ml predict housing.json -o predictions.json\fR + +Train a network to solve XOR problem, below is the training data: + + [ + { + "x": 0, + "y": 0, + "z": 0 + }, + { + "x": 1, + "y": 0, + "z": 1 + }, + { + "x": 0, + "y": 1, + "z": 1 + }, + { + "x": 1, + "y": 1, + "z": 0 + } + ] + +Use the default configuration file and train the model with: + $ ml train -e 150 xor.json + +And get the network output using: + $ ml predict xor.json .SH AUTHOR -Written by vech +Written by jvech .SH COPYRIGHT Copyright \(co 2023 jvech .PP |