llama-crab 0.1.2

Safe, ergonomic and complete Rust bindings for llama.cpp
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

//! `LlamaModelParams` builder.

use llama_crab_sys as sys;

/// Strategy for splitting a model across multiple GPUs.
#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash, Default)]
#[allow(dead_code)]
pub enum SplitMode {
    /// Single-GPU / no splitting.
    None,
    /// Split by layer (default).
    #[default]
    Layer,
    /// Split by row.
    Row,
    /// Split by tensor.
    Tensor,
}

impl From<SplitMode> for sys::llama_split_mode {
    fn from(s: SplitMode) -> Self {
        match s {
            SplitMode::None => Self::LLAMA_SPLIT_MODE_NONE,
            SplitMode::Layer => Self::LLAMA_SPLIT_MODE_LAYER,
            SplitMode::Row => Self::LLAMA_SPLIT_MODE_ROW,
            SplitMode::Tensor => Self::LLAMA_SPLIT_MODE_TENSOR,
        }
    }
}

/// Builder for `llama_model_params`.
///
/// `Default` produces the same parameters as `llama_model_default_params()`.
#[derive(Debug, Clone)]
pub struct LlamaModelParams {
    n_gpu_layers: i32,
    split_mode: SplitMode,
    main_gpu: i32,
    vocab_only: bool,
    use_mmap: bool,
    use_mlock: bool,
}

impl Default for LlamaModelParams {
    fn default() -> Self {
        let raw = unsafe { sys::llama_model_default_params() };
        Self {
            n_gpu_layers: raw.n_gpu_layers,
            split_mode: match raw.split_mode {
                sys::llama_split_mode::LLAMA_SPLIT_MODE_NONE => SplitMode::None,
                sys::llama_split_mode::LLAMA_SPLIT_MODE_LAYER => SplitMode::Layer,
                sys::llama_split_mode::LLAMA_SPLIT_MODE_ROW => SplitMode::Row,
                sys::llama_split_mode::LLAMA_SPLIT_MODE_TENSOR => SplitMode::Tensor,
                #[allow(unreachable_patterns)]
                _ => SplitMode::Layer,
            },
            main_gpu: raw.main_gpu,
            vocab_only: raw.vocab_only,
            use_mmap: raw.use_mmap,
            use_mlock: raw.use_mlock,
        }
    }
}

impl LlamaModelParams {
    /// Number of layers to offload to the GPU. `-1` offloads every layer.
    #[must_use]
    pub fn with_n_gpu_layers(mut self, n: i32) -> Self {
        self.n_gpu_layers = n;
        self
    }

    /// How to split a model across multiple GPUs.
    #[must_use]
    pub fn with_split_mode(mut self, mode: SplitMode) -> Self {
        self.split_mode = mode;
        self
    }

    /// Index of the main GPU.
    #[must_use]
    pub fn with_main_gpu(mut self, idx: i32) -> Self {
        self.main_gpu = idx;
        self
    }

    /// Load only the vocabulary (no model weights).
    #[must_use]
    pub fn with_vocab_only(mut self, yes: bool) -> Self {
        self.vocab_only = yes;
        self
    }

    /// Memory-map the model file.
    #[must_use]
    pub fn with_use_mmap(mut self, yes: bool) -> Self {
        self.use_mmap = yes;
        self
    }

    /// Lock the model into physical memory.
    #[must_use]
    pub fn with_use_mlock(mut self, yes: bool) -> Self {
        self.use_mlock = yes;
        self
    }

    /// Convert to the underlying C struct.
    #[allow(clippy::missing_const_for_fn)]
    pub(crate) fn build(&self) -> sys::llama_model_params {
        let mut raw = unsafe { sys::llama_model_default_params() };
        raw.n_gpu_layers = self.n_gpu_layers;
        raw.split_mode = self.split_mode.into();
        raw.main_gpu = self.main_gpu;
        raw.vocab_only = self.vocab_only;
        raw.use_mmap = self.use_mmap;
        raw.use_mlock = self.use_mlock;
        raw
    }
}