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
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
extern crate coaster as co;
extern crate coaster_nn as conn;
extern crate env_logger;
extern crate juice;
extern crate juice_utils;
extern crate mnist;

#[cfg(feature = "cuda")]
use co::frameworks::cuda::get_cuda_backend;
#[cfg(not(feature = "cuda"))]
use co::frameworks::native::get_native_backend;
use co::prelude::*;
use juice::layer::*;
use juice::layers::*;
use juice::solver::*;
use juice::util::*;
use juice_utils::{download_datasets, unzip_datasets};
use mnist::{Mnist, MnistBuilder};
use serde::Deserialize;
use std::rc::Rc;
use std::sync::{Arc, RwLock};

const MAIN_USAGE: &str = "
Juice Examples

Usage:
    juice-examples load-dataset <dataset-name>
    juice-examples mnist <model-name> [--batch-size <batch-size>] [--learning-rate <learning-rate>] [--momentum <momentum>]
    juice-examples fashion <model-name> [--batch-size <batch-size>] [--learning-rate <learning-rate>] [--momentum <momentum>]
    juice-examples (-h | --help)
    juice-examples --version


Options:
    <model-name>            Which MNIST model to use. Valid values: [linear, mlp, conv]
    <dataset-name>          Which dataset to download. Valid values: [mnist, fashion]

    -h --help               Show this screen.
    --version               Show version.
";

#[derive(Debug, Deserialize)]
struct Args {
    arg_dataset_name: Option<String>,
    arg_model_name: Option<String>,
    arg_batch_size: Option<usize>,
    arg_learning_rate: Option<f32>,
    arg_momentum: Option<f32>,
    cmd_load_dataset: bool,
    cmd_mnist: bool,
    cmd_fashion: bool,
}

enum MnistType {
    Fashion,
    Numbers,
}

#[cfg(not(test))]
#[allow(unused_must_use)]
fn main() {
    env_logger::init();
    // Parse Arguments
    let args: Args = docopt::Docopt::new(MAIN_USAGE)
        .and_then(|d| d.deserialize())
        .unwrap_or_else(|e| e.exit());

    if args.cmd_load_dataset {
        match &*args.arg_dataset_name.unwrap() {
            "mnist" => {
                let datasets = [
                    "train-images-idx3-ubyte.gz",
                    "train-labels-idx1-ubyte.gz",
                    "t10k-images-idx3-ubyte.gz",
                    "t10k-labels-idx1-ubyte.gz",
                ];
                download_datasets(
                    &datasets,
                    &"./assets/mnist/",
                    "http://yann.lecun.com/exdb/mnist/",
                )
                .unwrap();

                unzip_datasets(&datasets, &"./assets/mnist/").unwrap();
            }

            "fashion" => {
                let datasets = [
                    "train-images-idx3-ubyte.gz",
                    "train-labels-idx1-ubyte.gz",
                    "t10k-images-idx3-ubyte.gz",
                    "t10k-labels-idx1-ubyte.gz",
                ];

                download_datasets(
                    &datasets,
                    &"./assets/mnist-fashion/",
                    "http://fashion-mnist.s3-website.eu-central-1.amazonaws.com",
                )
                .unwrap();
                println!("{}", "Fashion MNIST dataset downloaded".to_string());

                unzip_datasets(&datasets, &"./assets/mnist-fashion/").unwrap();
                println!("{}", "Fashion MNIST dataset decompressed".to_string());
            }
            _ => println!("{}", "Failed to download MNIST dataset!".to_string()),
        }
    } else if args.cmd_mnist {
        #[cfg(all(feature = "cuda"))]
        run_mnist(
            MnistType::Numbers,
            args.arg_model_name,
            args.arg_batch_size,
            args.arg_learning_rate,
            args.arg_momentum,
        );
        #[cfg(not(feature = "cuda"))]
        {
            println!(
                    "Right now, you really need cuda! Not all features are available for all backends and as such, this one -as of now - only works with cuda."
                );
            panic!()
        }
    } else if args.cmd_fashion {
        run_mnist(
            MnistType::Fashion,
            args.arg_model_name,
            args.arg_batch_size,
            args.arg_learning_rate,
            args.arg_momentum,
        );
    }
}

#[cfg(all(feature = "cuda"))]
fn add_conv_net(
    mut net_cfg: SequentialConfig,
    batch_size: usize,
    pixel_dim: usize,
) -> SequentialConfig {
    net_cfg.add_layer(LayerConfig::new(
        "reshape",
        ReshapeConfig::of_shape(&[batch_size, 1, pixel_dim, pixel_dim]),
    ));
    net_cfg.add_layer(LayerConfig::new(
        "conv",
        ConvolutionConfig {
            num_output: 20,
            filter_shape: vec![5],
            padding: vec![0],
            stride: vec![1],
        },
    ));
    net_cfg.add_layer(LayerConfig::new(
        "pooling",
        PoolingConfig {
            mode: PoolingMode::Max,
            filter_shape: vec![2],
            padding: vec![0],
            stride: vec![2],
        },
    ));
    net_cfg.add_layer(LayerConfig::new(
        "linear1",
        LinearConfig { output_size: 500 },
    ));
    net_cfg.add_layer(LayerConfig::new("sigmoid", LayerType::Sigmoid));
    net_cfg.add_layer(LayerConfig::new(
        "linear2",
        LinearConfig { output_size: 10 },
    ));
    net_cfg
}

#[cfg(not(feature = "cuda"))]
fn add_conv_net(
    _net_cfg: SequentialConfig,
    _batch_size: usize,
    _pixel_dim: usize,
) -> SequentialConfig {
    println!(
        "Currently Juice does not have a native pooling function to use with Conv Nets - you can either try
        the CUDA implementation, or use a different type of layer"
    );
    panic!()
}

fn add_mlp(
    mut net_cfg: SequentialConfig,
    batch_size: usize,
    pixel_count: usize,
) -> SequentialConfig {
    net_cfg.add_layer(LayerConfig::new(
        "reshape",
        LayerType::Reshape(ReshapeConfig::of_shape(&[batch_size, pixel_count])),
    ));
    net_cfg.add_layer(LayerConfig::new(
        "linear1",
        LayerType::Linear(LinearConfig { output_size: 1568 }),
    ));
    net_cfg.add_layer(LayerConfig::new("sigmoid", LayerType::Sigmoid));
    net_cfg.add_layer(LayerConfig::new(
        "linear2",
        LayerType::Linear(LinearConfig { output_size: 10 }),
    ));
    net_cfg
}

fn add_linear_net(mut net_cfg: SequentialConfig) -> SequentialConfig {
    net_cfg.add_layer(LayerConfig::new(
        "linear",
        LayerType::Linear(LinearConfig { output_size: 10 }),
    ));
    net_cfg
}

fn run_mnist(
    mnist_type: MnistType,
    model_name: Option<String>,
    batch_size: Option<usize>,
    learning_rate: Option<f32>,
    momentum: Option<f32>,
) {
    let example_count = 60000;
    let test_count = 10000;
    let pixel_count = 784;
    let pixel_dim = 28;

    let asset_path = match mnist_type {
        MnistType::Fashion => "./assets/mnist-fashion",
        MnistType::Numbers => "./assets/mnist",
    };

    let Mnist {
        trn_img, trn_lbl, ..
    } = MnistBuilder::new()
        .base_path(asset_path)
        .label_format_digit()
        .training_set_length(example_count)
        .test_set_length(test_count)
        .finalize();

    let mut decoded_images = trn_img
        .chunks(pixel_count)
        .enumerate()
        .map(|(ind, pixels)| (trn_lbl[ind], pixels.to_vec()));

    let batch_size = batch_size.unwrap_or(30);
    let learning_rate = learning_rate.unwrap_or(0.001f32);
    let momentum = momentum.unwrap_or(0f32);

    let mut net_cfg = SequentialConfig::default();
    net_cfg.add_input("data", &[batch_size, pixel_dim, pixel_dim]);
    net_cfg.force_backward = true;

    net_cfg = match &*model_name.unwrap_or("none".to_owned()) {
        "conv" => add_conv_net(net_cfg, batch_size, pixel_dim),
        "mlp" => add_mlp(net_cfg, batch_size, pixel_count),
        "linear" => add_linear_net(net_cfg),
        _ => panic!("Unknown model. Try one of [linear, mlp, conv]"),
    };

    net_cfg.add_layer(LayerConfig::new("log_softmax", LayerType::LogSoftmax));

    let mut classifier_cfg = SequentialConfig::default();
    classifier_cfg.add_input("network_out", &[batch_size, 10]);
    classifier_cfg.add_input("label", &[batch_size, 1]);
    // set up nll loss
    let nll_layer_cfg = NegativeLogLikelihoodConfig { num_classes: 10 };
    let nll_cfg = LayerConfig::new("nll", LayerType::NegativeLogLikelihood(nll_layer_cfg));
    classifier_cfg.add_layer(nll_cfg);

    // set up backends
    #[cfg(all(feature = "cuda"))]
    let backend = Rc::new(get_cuda_backend());
    #[cfg(not(feature = "cuda"))]
    let backend = Rc::new(get_native_backend());

    // set up solver
    let mut solver_cfg = SolverConfig {
        minibatch_size: batch_size,
        base_lr: learning_rate,
        momentum,
        ..SolverConfig::default()
    };
    solver_cfg.network = LayerConfig::new("network", net_cfg);
    solver_cfg.objective = LayerConfig::new("classifier", classifier_cfg);
    let mut solver = Solver::from_config(backend.clone(), backend.clone(), &solver_cfg);

    // set up confusion matrix
    let mut classification_evaluator = ::juice::solver::ConfusionMatrix::new(10);
    classification_evaluator.set_capacity(Some(1000));

    let input = SharedTensor::<f32>::new(&[batch_size, pixel_dim, pixel_dim]);
    let inp_lock = Arc::new(RwLock::new(input));

    let label = SharedTensor::<f32>::new(&[batch_size, 1]);
    let label_lock = Arc::new(RwLock::new(label));

    for _ in 0..(example_count / batch_size as u32) {
        // write input
        let mut targets = Vec::new();

        for (batch_n, (label_val, ref input)) in
            decoded_images.by_ref().take(batch_size).enumerate()
        {
            let mut input_tensor = inp_lock.write().unwrap();
            let mut label_tensor = label_lock.write().unwrap();
            write_batch_sample(&mut input_tensor, &input, batch_n);
            write_batch_sample(&mut label_tensor, &[label_val], batch_n);
            targets.push(label_val as usize);
        }
        // train the network!
        let infered_out = solver.train_minibatch(inp_lock.clone(), label_lock.clone());

        let mut infered = infered_out.write().unwrap();
        let predictions = classification_evaluator.get_predictions(&mut infered);

        classification_evaluator.add_samples(&predictions, &targets);
        println!(
            "Last sample: {} | Accuracy {}",
            classification_evaluator.samples().iter().last().unwrap(),
            classification_evaluator.accuracy()
        );
    }
}