aprender-serve 0.64.0

Pure Rust ML inference engine built from scratch - model serving for GGUF and safetensors
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


#[test]
fn test_vocabulary_single_token() {
    let data = GGUFBuilder::new()
        .architecture("llama")
        .add_string_array("tokenizer.ggml.tokens", &["<unk>"])
        .build();
    let model = GGUFModel::from_bytes(&data).expect("parse");
    let vocab = model.vocabulary();
    assert!(vocab.is_some());
    assert_eq!(vocab.expect("vocab").len(), 1);
}

// ============================================================================
// GGUFConfig edge cases
// ============================================================================

#[test]
fn test_gguf_config_from_gguf_with_rope_freq_base() {
    use crate::gguf::test_factory::build_minimal_llama_gguf;
    let data = build_minimal_llama_gguf(100, 64, 256, 4, 4);
    let model = GGUFModel::from_bytes(&data).expect("parse");
    let config = crate::gguf::GGUFConfig::from_gguf(&model);
    assert!(config.is_ok());
    let config = config.expect("config");
    assert_eq!(config.hidden_dim, 64);
    assert_eq!(config.num_heads, 4);
    assert!(config.rope_theta > 0.0);
}

// ============================================================================
// Decode mixed tokens (byte + regular + unknown)
// ============================================================================

#[test]
fn test_decode_mixed_byte_and_regular_tokens() {
    let data = GGUFBuilder::new()
        .architecture("llama")
        .add_string_array(
            "tokenizer.ggml.tokens",
            &["Hello", "<0x20>", "world", "<0x21>"],
        )
        .build();
    let model = GGUFModel::from_bytes(&data).expect("parse");
    // Hello<space>world!
    let decoded = model.decode(&[0, 1, 2, 3]);
    assert_eq!(decoded, "Hello world!");
}

#[test]
fn test_decode_invalid_byte_token_format() {
    // <0xZZ> is not valid hex, should NOT be treated as byte token
    let data = GGUFBuilder::new()
        .architecture("llama")
        .add_string_array("tokenizer.ggml.tokens", &["<0xZZ>"])
        .build();
    let model = GGUFModel::from_bytes(&data).expect("parse");
    let decoded = model.decode(&[0]);
    // Should treat as regular string token, not byte
    assert!(decoded.contains("<0xZZ>"));
}