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

#[cfg(test)]
#[allow(clippy::module_inception)]
mod tests {
    use crate::llama::config::LlamaConfig;
    use crate::llama::model::{LlamaForCausalLM, LlamaModel, RMSNorm, RotaryEmbedding};
    use trustformers_core::traits::Config;

    #[test]
    fn test_llama_config_validation() {
        // Use minimal config to reduce memory usage
        let config = LlamaConfig {
            num_hidden_layers: 2,
            vocab_size: 1000,
            hidden_size: 64,
            num_attention_heads: 8,
            intermediate_size: 256,
            ..LlamaConfig::default()
        };
        assert!(config.validate().is_ok());

        // Test head dimension calculation
        assert_eq!(config.head_dim(), 8); // 64 / 8

        // Test KV heads
        assert_eq!(config.num_kv_heads(), 8); // No grouped-query attention by default
        assert_eq!(config.num_query_groups(), 1);

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

    #[test]
    fn test_llama2_config_with_gqa() {
        // Use minimal config with GQA to reduce memory usage
        let config = LlamaConfig {
            num_hidden_layers: 2,
            vocab_size: 1000,
            hidden_size: 64,
            num_attention_heads: 8,
            num_key_value_heads: Some(2), // Grouped-query attention
            intermediate_size: 256,
            ..LlamaConfig::default()
        };
        assert!(config.validate().is_ok());

        // Test grouped-query attention
        assert_eq!(config.num_kv_heads(), 2);
        assert_eq!(config.num_query_groups(), 4); // 8 / 2
        assert_eq!(config.head_dim(), 8); // 64 / 8

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

    #[test]
    fn test_code_llama_config() {
        // Use minimal config inspired by Code Llama to reduce memory usage
        let config = LlamaConfig {
            num_hidden_layers: 2,
            vocab_size: 1000,
            hidden_size: 64,
            num_attention_heads: 8,
            intermediate_size: 256,
            max_position_embeddings: 512, // Reduced from 16384
            ..LlamaConfig::default()
        };
        assert!(config.validate().is_ok());
        assert_eq!(config.max_position_embeddings, 512); // Reduced context for testing
        assert_eq!(config.vocab_size, 1000); // Smaller vocab

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

    #[test]
    fn test_llama_architecture() {
        let config = LlamaConfig::default();
        assert_eq!(config.architecture(), "LLaMA");
    }

    #[test]
    fn test_invalid_llama_config() {
        let config = LlamaConfig {
            hidden_size: 4095, // Not divisible by num_attention_heads (32)
            ..LlamaConfig::default()
        };
        assert!(config.validate().is_err());
    }

    #[test]
    fn test_rmsnorm_creation() {
        let rmsnorm = RMSNorm::new(4096, 1e-6);
        assert!(rmsnorm.is_ok());
    }

    #[test]
    fn test_rotary_embedding_creation() {
        let rope = RotaryEmbedding::new(128, 2048, 10000.0);
        // Just test that it doesn't panic
        assert_eq!(rope.dim, 128);
        assert_eq!(rope.max_seq_len, 2048);
        assert_eq!(rope.base, 10000.0);
    }

    #[test]
    fn test_llama_model_creation() {
        let config = LlamaConfig {
            num_hidden_layers: 2, // Smaller for testing
            vocab_size: 1000,
            hidden_size: 64,
            num_attention_heads: 8,
            intermediate_size: 256,
            ..LlamaConfig::default()
        };

        let model = LlamaModel::new(config);
        assert!(model.is_ok());
    }

    #[test]
    fn test_llama_for_causal_lm_creation() {
        let config = LlamaConfig {
            num_hidden_layers: 1, // Minimal for testing
            vocab_size: 100,
            hidden_size: 32,
            num_attention_heads: 4,
            intermediate_size: 128,
            ..LlamaConfig::default()
        };

        let model = LlamaForCausalLM::new(config);
        assert!(model.is_ok());
    }
}