oxionnx 0.1.3

Pure Rust ONNX inference engine — zero C/C++ dependencies
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

//! Shared helpers for operator integration tests.
#![allow(dead_code)]

use std::collections::HashMap;

use oxionnx::{Attributes, Graph, Node, OpKind, OptLevel, Session, Tensor};

pub fn make_node(op: OpKind, name: &str, inputs: &[&str], outputs: &[&str]) -> Node {
    Node {
        op,
        name: name.to_string(),
        inputs: inputs.iter().map(|s| s.to_string()).collect(),
        outputs: outputs.iter().map(|s| s.to_string()).collect(),
        attrs: Attributes::default(),
    }
}

pub fn make_node_with_attrs(
    op: OpKind,
    name: &str,
    inputs: &[&str],
    outputs: &[&str],
    attrs: Attributes,
) -> Node {
    Node {
        op,
        name: name.to_string(),
        inputs: inputs.iter().map(|s| s.to_string()).collect(),
        outputs: outputs.iter().map(|s| s.to_string()).collect(),
        attrs,
    }
}

/// Run a single-op graph and return outputs.
pub fn run_single_op(
    op: OpKind,
    inputs: Vec<(&str, Tensor)>,
    weights: Vec<(&str, Tensor)>,
    input_names: Vec<&str>,
    node_inputs: Vec<&str>,
    node_output: &str,
    attrs: Attributes,
) -> HashMap<String, Tensor> {
    let node = make_node_with_attrs(op, "op0", &node_inputs, &[node_output], attrs);
    let graph = Graph {
        nodes: vec![node],
        input_names: input_names.iter().map(|s| s.to_string()).collect(),
        output_names: vec![node_output.to_string()],
        ..Default::default()
    };
    let mut w: HashMap<String, Tensor> = HashMap::new();
    for (name, tensor) in weights {
        w.insert(name.to_string(), tensor);
    }
    let session = Session::builder()
        .with_optimization_level(OptLevel::None)
        .build_from_graph(graph, w)
        .expect("build session");
    let mut feed: HashMap<&str, Tensor> = HashMap::new();
    for (name, tensor) in inputs {
        feed.insert(name, tensor);
    }
    session.run(&feed).expect("run")
}

/// Run a single-op graph with multiple outputs.
pub fn run_single_op_multi_output(
    op: OpKind,
    inputs: Vec<(&str, Tensor)>,
    weights: Vec<(&str, Tensor)>,
    input_names: Vec<&str>,
    node_inputs: Vec<&str>,
    node_outputs: Vec<&str>,
    attrs: Attributes,
) -> HashMap<String, Tensor> {
    let node = make_node_with_attrs(op, "op0", &node_inputs, &node_outputs, attrs);
    let graph = Graph {
        nodes: vec![node],
        input_names: input_names.iter().map(|s| s.to_string()).collect(),
        output_names: node_outputs.iter().map(|s| s.to_string()).collect(),
        ..Default::default()
    };
    let mut w: HashMap<String, Tensor> = HashMap::new();
    for (name, tensor) in weights {
        w.insert(name.to_string(), tensor);
    }
    let session = Session::builder()
        .with_optimization_level(OptLevel::None)
        .build_from_graph(graph, w)
        .expect("build session");
    let mut feed: HashMap<&str, Tensor> = HashMap::new();
    for (name, tensor) in inputs {
        feed.insert(name, tensor);
    }
    session.run(&feed).expect("run")
}

/// Run a single-op graph (conformance-style: separate graph_inputs and node_outputs).
pub fn run_op(
    op: OpKind,
    node_inputs: Vec<&str>,
    node_outputs: Vec<&str>,
    graph_inputs: Vec<&str>,
    input_tensors: Vec<(&str, Tensor)>,
    weights: Vec<(&str, Tensor)>,
    attrs: Attributes,
) -> HashMap<String, Tensor> {
    let node = make_node_with_attrs(op, "op0", &node_inputs, &node_outputs, attrs);
    let graph = Graph {
        nodes: vec![node],
        input_names: graph_inputs.iter().map(|s| s.to_string()).collect(),
        output_names: node_outputs.iter().map(|s| s.to_string()).collect(),
        ..Default::default()
    };
    let mut w: HashMap<String, Tensor> = HashMap::new();
    for (name, tensor) in weights {
        w.insert(name.to_string(), tensor);
    }
    let session = Session::builder()
        .with_optimization_level(OptLevel::None)
        .build_from_graph(graph, w)
        .expect("build session");
    let mut feed: HashMap<&str, Tensor> = HashMap::new();
    for (name, tensor) in input_tensors {
        feed.insert(name, tensor);
    }
    session.run(&feed).expect("run")
}

pub fn assert_close(actual: &[f32], expected: &[f32], tol: f32, msg: &str) {
    assert_eq!(
        actual.len(),
        expected.len(),
        "{}: length mismatch (got {} expected {})",
        msg,
        actual.len(),
        expected.len()
    );
    for (i, (a, e)) in actual.iter().zip(expected).enumerate() {
        assert!(
            (a - e).abs() <= tol,
            "{}: idx {} got {} expected {} (tol {})",
            msg,
            i,
            a,
            e,
            tol
        );
    }
}

pub fn assert_shape(tensor: &Tensor, expected: &[usize], msg: &str) {
    assert_eq!(tensor.shape, expected, "{}: shape mismatch", msg);
}

pub fn assert_tensor_approx(actual: &Tensor, expected: &[f32], tol: f32) {
    assert_eq!(
        actual.data.len(),
        expected.len(),
        "length mismatch: got {} expected {}",
        actual.data.len(),
        expected.len()
    );
    for (i, (a, e)) in actual.data.iter().zip(expected).enumerate() {
        assert!(
            (a - e).abs() < tol,
            "index {}: {} vs {} (tol={})",
            i,
            a,
            e,
            tol
        );
    }
}