modelc 0.1.9

Rust CLI that compiles LLM weights (GGUF, Safetensors, ONNX, PyTorch) into a single .modelc artifact and serves a local OpenAI-compatible inference API with Metal GPU and CPU SIMD acceleration.
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

use std::collections::HashMap;

use anyhow::{Result, ensure};

use crate::runtime::tensor::Tensor;

use super::AttributeValue;

pub(super) fn get_attr_f32(
    attrs: &HashMap<String, AttributeValue>,
    key: &str,
    default: f32,
) -> f32 {
    attrs
        .get(key)
        .map(|v| match v {
            AttributeValue::Float(f) => *f,
            AttributeValue::Int(i) => *i as f32,
            _ => default,
        })
        .unwrap_or(default)
}

pub(super) fn get_attr_int(
    attrs: &HashMap<String, AttributeValue>,
    key: &str,
    default: i64,
) -> i64 {
    attrs
        .get(key)
        .map(|v| match v {
            AttributeValue::Int(i) => *i,
            AttributeValue::Float(f) => *f as i64,
            _ => default,
        })
        .unwrap_or(default)
}

pub(super) fn get_attr_ints(
    attrs: &HashMap<String, AttributeValue>,
    key: &str,
) -> Option<Vec<i64>> {
    attrs.get(key).map(|v| match v {
        AttributeValue::Ints(v) => v.clone(),
        _ => Vec::new(),
    })
}

pub(super) fn transpose(a: &Tensor) -> Tensor {
    if a.shape.len() != 2 {
        return Tensor::from_vec(a.data.clone(), a.shape.clone());
    }
    let (m, n) = (a.shape[0], a.shape[1]);
    let mut out = vec![0.0f32; m * n];
    for i in 0..m {
        for j in 0..n {
            out[j * m + i] = a.data[i * n + j];
        }
    }
    Tensor::from_vec(out, vec![n, m])
}

pub(super) fn transpose_with_perm(a: &Tensor, perm: &[i64]) -> Result<Tensor> {
    ensure!(
        perm.len() == a.shape.len(),
        "transpose perm length mismatch"
    );
    let new_shape: Vec<usize> = perm.iter().map(|&p| a.shape[p as usize]).collect();
    let mut out = vec![0.0f32; a.data.len()];

    let rank = a.shape.len();
    let mut strides = vec![1usize; rank];
    for i in (0..rank - 1).rev() {
        strides[i] = strides[i + 1] * a.shape[i + 1];
    }
    let mut new_strides = vec![1usize; rank];
    for i in (0..rank - 1).rev() {
        new_strides[i] = new_strides[i + 1] * new_shape[i + 1];
    }

    for idx in 0..a.data.len() {
        let mut coords = vec![0usize; rank];
        let mut rem = idx;
        for i in 0..rank {
            coords[i] = rem / strides[i];
            rem %= strides[i];
        }
        let new_idx: usize = perm
            .iter()
            .enumerate()
            .map(|(i, &p)| coords[p as usize] * new_strides[i])
            .sum();
        out[new_idx] = a.data[idx];
    }

    Ok(Tensor::from_vec(out, new_shape))
}

pub(super) fn element_wise_mul(a: &Tensor, b: &Tensor) -> Result<Tensor> {
    if a.shape == b.shape {
        let data: Vec<f32> = a.data.iter().zip(&b.data).map(|(x, y)| x * y).collect();
        Ok(Tensor::from_vec(data, a.shape.clone()))
    } else if b.data.len() == 1 {
        Ok(crate::runtime::ops::mul_scalar(a, b.data[0]))
    } else if a.data.len() == 1 {
        Ok(crate::runtime::ops::mul_scalar(b, a.data[0]))
    } else {
        anyhow::bail!("Mul: incompatible shapes {:?} and {:?}", a.shape, b.shape)
    }
}

pub(super) fn element_wise_div(a: &Tensor, b: &Tensor) -> Result<Tensor> {
    if a.shape == b.shape {
        let data: Vec<f32> = a.data.iter().zip(&b.data).map(|(x, y)| x / y).collect();
        Ok(Tensor::from_vec(data, a.shape.clone()))
    } else if b.data.len() == 1 {
        let s = b.data[0];
        let data: Vec<f32> = a.data.iter().map(|x| x / s).collect();
        Ok(Tensor::from_vec(data, a.shape.clone()))
    } else {
        anyhow::bail!("Div: incompatible shapes {:?} and {:?}", a.shape, b.shape)
    }
}

pub(super) fn element_wise_sub(a: &Tensor, b: &Tensor) -> Result<Tensor> {
    if a.shape == b.shape {
        let data: Vec<f32> = a.data.iter().zip(&b.data).map(|(x, y)| x - y).collect();
        Ok(Tensor::from_vec(data, a.shape.clone()))
    } else {
        anyhow::bail!("Sub: incompatible shapes {:?} and {:?}", a.shape, b.shape)
    }
}

pub(super) fn sigmoid(a: &Tensor) -> Tensor {
    let data: Vec<f32> = a.data.iter().map(|x| 1.0 / (1.0 + (-x).exp())).collect();
    Tensor::from_vec(data, a.shape.clone())
}

pub(super) fn tanh_tensor(a: &Tensor) -> Tensor {
    let data: Vec<f32> = a.data.iter().map(|x| x.tanh()).collect();
    Tensor::from_vec(data, a.shape.clone())
}