libffm 0.2.0

Field-aware factorization machines in Rust
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174

use crate::disk::{ProblemLoader, ProblemOnDisk};
use crate::error::Error;
use crate::model::Model;
use crate::timer::Timer;
use std::io::{Read, Seek, Write};
use std::path::Path;

/// A set of parameters.
#[derive(Debug)]
pub struct Params {
    pub(crate) eta: f32,
    pub(crate) lambda: f32,
    nr_iters: i32,
    pub(crate) k: i32,
    pub(crate) normalization: bool,
    auto_stop: bool,
    quiet: bool,
    on_disk: bool,
}

impl Params {
    /// Returns a new set of parameters.
    pub fn new() -> Self {
        Self {
            eta: 0.2,
            lambda: 0.00002,
            nr_iters: 15,
            k: 4,
            normalization: true,
            auto_stop: false,
            quiet: false,
            on_disk: false,
        }
    }

    /// Sets the learning rate.
    pub fn learning_rate(&mut self, value: f32) -> &mut Self {
        self.eta = value;
        self
    }

    /// Sets the regularization parameter.
    pub fn lambda(&mut self, value: f32) -> &mut Self {
        self.lambda = value;
        self
    }

    /// Sets the number of iterations.
    pub fn iterations(&mut self, value: i32) -> &mut Self {
        self.nr_iters = value;
        self
    }

    /// Sets the number of latent factors.
    pub fn factors(&mut self, value: i32) -> &mut Self {
        self.k = value;
        self
    }

    /// Sets whether to use instance-wise normalization.
    pub fn normalization(&mut self, value: bool) -> &mut Self {
        self.normalization = value;
        self
    }

    /// Sets whether to stop at the iteration that achieves the best validation loss.
    pub fn auto_stop(&mut self, value: bool) -> &mut Self {
        self.auto_stop = value;
        self
    }

    /// Sets whether to use quiet mode (no output).
    pub fn quiet(&mut self, value: bool) -> &mut Self {
        self.quiet = value;
        self
    }

    /// Sets whether to use on-disk training.
    pub fn on_disk(&mut self, value: bool) -> &mut Self {
        self.on_disk = value;
        self
    }

    /// Trains a model.
    pub fn train<P: AsRef<Path>>(&self, tr_path: P) -> Result<Model, Error> {
        let mut tr_loader = ProblemLoader::new(tr_path, self.quiet)?;

        if self.on_disk {
            let tr_problem = tr_loader.read_to_disk()?;
            self.train_core(tr_problem)
        } else {
            let tr_problem = tr_loader.read_to_memory()?;
            self.train_core(tr_problem)
        }
    }

    /// Trains a model and performs cross-validation.
    pub fn train_eval<P: AsRef<Path>, Q: AsRef<Path>>(&self, tr_path: P, va_path: Q) -> Result<Model, Error> {
        // open both files first to fail fast
        let mut tr_loader = ProblemLoader::new(tr_path, self.quiet)?;
        let mut va_loader = ProblemLoader::new(va_path, self.quiet)?;

        if self.on_disk {
            let tr_problem = tr_loader.read_to_disk()?;
            let va_problem = va_loader.read_to_disk()?;
            self.train_eval_core(tr_problem, va_problem)
        } else {
            let tr_problem = tr_loader.read_to_memory()?;
            let va_problem = va_loader.read_to_memory()?;
            self.train_eval_core(tr_problem, va_problem)
        }
    }

    fn train_core<W: Read + Write + Seek>(&self, mut tr: ProblemOnDisk<W>) -> Result<Model, Error> {
        let mut model = Model::new(tr.meta.n, tr.meta.m, self);

        self.logln(format!("{:>4}{:>13}{:>13}", "iter", "tr_logloss", "tr_time"));

        let mut timer = Timer::new();

        for iter in 1..=self.nr_iters {
            timer.tic();
            let tr_loss = model.one_epoch(&mut tr, true, self)?;
            timer.toc();
            self.logln(format!("{:>4}{:>13.5}{:>13.1}", iter, tr_loss, timer.get()));
        }

        Ok(model)
    }

    fn train_eval_core<W: Read + Write + Seek>(&self, mut tr: ProblemOnDisk<W>, mut va: ProblemOnDisk<W>) -> Result<Model, Error> {
        let mut model = Model::new(tr.meta.n, tr.meta.m, self);
        let mut prev_w = Vec::new();
        let mut best_va_loss = f32::MAX;

        self.logln(format!("{:>4}{:>13}{:>13}{:>13}", "iter", "tr_logloss", "va_logloss", "tr_time"));

        let mut timer = Timer::new();

        for iter in 1..=self.nr_iters {
            timer.tic();
            let tr_loss = model.one_epoch(&mut tr, true, self)?;
            timer.toc();

            let va_loss = model.one_epoch(&mut va, false, self)?;
            if self.auto_stop {
                if va_loss > best_va_loss {
                    model.w = prev_w;
                    self.logln(format!("Auto-stop. Use model at {}th iteration.", iter - 1));
                    break;
                } else {
                    prev_w.clone_from(&model.w);
                    best_va_loss = va_loss;
                }
            }

            self.logln(format!("{:>4}{:>13.5}{:>13.5}{:>13.1}", iter, tr_loss, va_loss, timer.get()));
        }

        Ok(model)
    }

    fn logln(&self, msg: String) {
        if !self.quiet {
            println!("{}", msg);
        }
    }
}

impl Default for Params {
    fn default() -> Self {
        Self::new()
    }
}