trustformers-models 0.1.1

Model implementations for TrustformeRS
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
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

#[cfg(test)]
#[allow(clippy::module_inception)]
mod tests {
    use crate::vit::config::ViTConfig;
    use crate::vit::model::{PatchEmbedding, ViTForImageClassification, ViTModel};
    use scirs2_core::ndarray::Array4; // SciRS2 Integration Policy
    use trustformers_core::traits::Config;

    #[test]
    fn test_vit_config() {
        let config = ViTConfig::base();
        assert_eq!(config.image_size, 224);
        assert_eq!(config.patch_size, 16);
        assert_eq!(config.num_patches(), 196); // (224/16)^2
        assert_eq!(config.seq_length(), 197); // 196 patches + 1 class token

        config.validate().expect("operation failed");
    }

    #[test]
    fn test_vit_config_variants() {
        let tiny = ViTConfig::tiny();
        assert_eq!(tiny.hidden_size, 192);
        assert_eq!(tiny.num_attention_heads, 3);
        assert_eq!(tiny.num_hidden_layers, 12);

        let small = ViTConfig::small();
        assert_eq!(small.hidden_size, 384);
        assert_eq!(small.num_attention_heads, 6);
        assert_eq!(small.num_hidden_layers, 12);

        let large = ViTConfig::large();
        assert_eq!(large.hidden_size, 1024);
        assert_eq!(large.num_attention_heads, 16);
        assert_eq!(large.num_hidden_layers, 24);

        let huge = ViTConfig::huge();
        assert_eq!(huge.hidden_size, 1280);
        assert_eq!(huge.num_attention_heads, 16);
        assert_eq!(huge.num_hidden_layers, 32);
        assert_eq!(huge.patch_size, 14); // Huge uses 14x14 patches
    }

    #[test]
    fn test_patch_embedding() {
        // Use minimal config to reduce memory usage
        let config = ViTConfig {
            image_size: 32,
            patch_size: 16,
            hidden_size: 64,
            num_attention_heads: 4,
            intermediate_size: 256,
            num_hidden_layers: 2,
            ..ViTConfig::default()
        };
        let patch_embedding = PatchEmbedding::new(&config);

        // Test with a small 32x32x3 image to reduce memory usage
        let image = Array4::zeros((1, 32, 32, 3));
        let result = patch_embedding.forward(&image);

        assert!(result.is_ok());
        let patches = result.expect("operation failed");
        assert_eq!(patches.shape(), &[1, 4, 64]); // 1 batch, 4 patches (32/16)^2, 64 hidden

        // Explicit cleanup
        drop(patches);
        drop(patch_embedding);
        std::hint::black_box(());
    }

    #[test]
    fn test_patch_embedding_different_sizes() {
        // Use minimal config to reduce memory usage
        let mut config = ViTConfig {
            image_size: 32,
            patch_size: 16,
            hidden_size: 64,
            num_attention_heads: 4,
            intermediate_size: 256,
            num_hidden_layers: 2,
            ..ViTConfig::default()
        };
        config.patch_size = 16;
        config.image_size = 32;

        let patch_embedding = PatchEmbedding::new(&config);

        // Test with small 32x32x3 image and 16x16 patches to reduce memory usage
        let image = Array4::zeros((1, 32, 32, 3));
        let result = patch_embedding.forward(&image).expect("operation failed");

        let expected_patches = (32 / 16) * (32 / 16); // 2x2 = 4 patches
        assert_eq!(result.shape(), &[1, expected_patches, 64]);

        // Explicit cleanup
        drop(result);
        drop(patch_embedding);
        std::hint::black_box(());
    }

    #[test]
    fn test_vit_model() {
        // Use even smaller config to reduce memory usage
        let config = ViTConfig {
            image_size: 32,
            patch_size: 16,
            hidden_size: 64,
            num_attention_heads: 4,
            intermediate_size: 256,
            num_hidden_layers: 2,
            ..ViTConfig::default()
        };
        let model = ViTModel::new(config).expect("operation failed");

        // Test with a single small image to reduce memory usage
        let images = Array4::zeros((1, 32, 32, 3));
        let result = model.forward(&images);

        assert!(result.is_ok());
        let output = result.expect("operation failed");
        assert_eq!(output.shape(), &[1, 5, 64]); // batch, seq_len (4 patches + 1 class token), hidden_size

        // Explicit cleanup
        drop(output);
        drop(model);
        std::hint::black_box(());
    }

    #[test]
    fn test_vit_classification() {
        // Use smaller config and batch size to reduce memory usage
        let config = ViTConfig {
            image_size: 32,
            patch_size: 16,
            hidden_size: 64,
            num_attention_heads: 4,
            intermediate_size: 256,
            num_hidden_layers: 2,
            num_labels: 10, // Reduced from 1000
            ..ViTConfig::default()
        };
        let model = ViTForImageClassification::new(config).expect("operation failed");

        let images = Array4::zeros((1, 32, 32, 3));
        let result = model.forward(&images);

        assert!(result.is_ok());
        let logits = result.expect("operation failed");
        assert_eq!(logits.shape(), &[1, 10]); // batch_size, num_classes

        // Explicit cleanup
        drop(logits);
        drop(model);
        std::hint::black_box(());
    }

    #[test]
    fn test_vit_class_token_output() {
        // Use smaller config to reduce memory usage
        let config = ViTConfig {
            image_size: 32,
            patch_size: 16,
            hidden_size: 64,
            num_attention_heads: 4,
            intermediate_size: 256,
            num_hidden_layers: 2,
            ..ViTConfig::default()
        };
        let model = ViTModel::new(config).expect("operation failed");

        let images = Array4::zeros((1, 32, 32, 3));
        let result = model.get_class_token_output(&images);

        assert!(result.is_ok());
        let class_output = result.expect("operation failed");
        assert_eq!(class_output.shape(), &[1, 64]); // batch_size, hidden_size

        // Explicit cleanup
        drop(class_output);
        drop(model);
        std::hint::black_box(());
    }

    #[test]
    fn test_vit_without_class_token() {
        // Use smaller config to reduce memory usage
        let mut config = ViTConfig {
            image_size: 32,
            patch_size: 16,
            hidden_size: 64,
            num_attention_heads: 4,
            intermediate_size: 256,
            num_hidden_layers: 2,
            ..ViTConfig::default()
        };
        config.use_class_token = false;

        let model = ViTModel::new(config).expect("operation failed");

        let images = Array4::zeros((1, 32, 32, 3));
        let output = model.forward(&images).expect("operation failed");

        // Should have 4 patches (no class token) (32/16)^2
        assert_eq!(output.shape(), &[1, 4, 64]);

        // Class token output should be mean of patches
        let class_output = model.get_class_token_output(&images).expect("operation failed");
        assert_eq!(class_output.shape(), &[1, 64]);

        // Explicit cleanup
        drop(output);
        drop(class_output);
        drop(model);
        std::hint::black_box(());
    }

    #[test]
    fn test_from_pretrained_name() {
        let base = ViTConfig::from_pretrained_name("vit-base-patch16-224");
        assert_eq!(base.hidden_size, 768);

        let large = ViTConfig::from_pretrained_name("vit-large-patch16-224");
        assert_eq!(large.hidden_size, 1024);

        let tiny = ViTConfig::from_pretrained_name("vit-tiny-patch16-224");
        assert_eq!(tiny.hidden_size, 192);
    }

    #[test]
    fn test_config_validation_errors() {
        let mut config = ViTConfig::base();

        // Test invalid hidden_size / num_attention_heads
        config.hidden_size = 100;
        config.num_attention_heads = 12;
        assert!(config.validate().is_err());

        // Test invalid image_size / patch_size
        config = ViTConfig::base();
        config.image_size = 225; // Not divisible by 16
        assert!(config.validate().is_err());

        // Test zero patch_size
        config = ViTConfig::base();
        config.patch_size = 0;
        assert!(config.validate().is_err());
    }

    #[test]
    fn test_config_with_different_patch_sizes() {
        let base = ViTConfig::base();
        let patch32 = base.with_patch_size(32);

        assert_eq!(patch32.patch_size, 32);
        assert_eq!(patch32.encoder_stride, 32);
        assert_eq!(patch32.num_patches(), 49); // (224/32)^2
    }

    #[test]
    fn test_config_with_different_image_sizes() {
        let base = ViTConfig::base();
        let img384 = base.with_image_size(384);

        assert_eq!(img384.image_size, 384);
        assert_eq!(img384.num_patches(), 576); // (384/16)^2
    }
}