llm_interface 0.0.3

llm_interface: The backend for the llm_client crate
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
175
176
177
178
179
180
181
182
183
184
185
186
187

use super::*;
use llm_interface::llms::local::llama_cpp::server::{get_all_server_pids, kill_server_from_model};

#[tokio::test]
#[serial]
async fn test_dropping_server() {
    let loaded = LlmInterface::llama_cpp().init().await.unwrap();
    std::mem::drop(loaded);
    let pids = get_all_server_pids().unwrap();
    assert!(pids.is_empty());
}

#[tokio::test]
#[serial]
async fn test_killing_server() {
    let loaded = LlmInterface::llama_cpp().init().await.unwrap();

    loaded.shutdown();

    let pids = get_all_server_pids().unwrap();
    assert!(pids.is_empty());
}

#[tokio::test]
#[serial]
async fn test_killing_server_with_model_id() {
    let loaded = LlmInterface::llama_cpp().init().await.unwrap();

    let model_id = loaded.model_id();
    kill_server_from_model(&model_id).unwrap();

    let pids = get_all_server_pids().unwrap();
    assert!(pids.is_empty());
}

#[tokio::test]
#[serial]
async fn test_multiple_servers() {
    let loaded_1 = LlmInterface::llama_cpp().init().await.unwrap();
    let _loaded_2 = LlmInterface::llama_cpp()
        .with_api_port("8081")
        .init()
        .await
        .unwrap();
    let pids = get_all_server_pids().unwrap();
    assert_eq!(pids.len(), 2);
    loaded_1.shutdown();
    let pids = get_all_server_pids().unwrap();
    assert_eq!(pids.len(), 1);
}

#[tokio::test]
#[serial]
async fn test_auto_gpu() {
    let backend = LlmInterface::llama_cpp().init().await.unwrap();
    assert!(
        backend
            .llama_cpp()
            .unwrap()
            .server
            .device_config
            .gpu_count()
            > 0
    );
    let mut req = CompletionRequest::new(backend);
    req.prompt
        .add_user_message()
        .unwrap()
        .set_content("Hello, world!");

    let res = req.request().await.unwrap();
    println!("{res}");
}

#[cfg(any(target_os = "linux", target_os = "windows"))]
#[tokio::test]
#[serial]
async fn test_single_gpu_map() {
    use llm_interface::llms::local::LlmLocalTrait;

    let cuda_config = CudaConfig::new_from_cuda_devices(vec![0]);

    let backend = LlmInterface::llama_cpp()
        .cuda_config(cuda_config)
        .init()
        .await
        .unwrap();
    assert!(
        backend
            .llama_cpp()
            .unwrap()
            .server
            .device_config
            .gpu_count()
            == 1
    );
    let mut req = CompletionRequest::new(backend);
    req.prompt
        .add_user_message()
        .unwrap()
        .set_content("Hello, world!");

    let res = req.request().await.unwrap();
    println!("{res}");
}

#[cfg(any(target_os = "linux", target_os = "windows"))]
#[tokio::test]
#[serial]
async fn test_two_gpu_map() {
    let cuda_config = CudaConfig::new_from_cuda_devices(vec![0, 1]);

    let backend = LlmInterface::llama_cpp()
        .cuda_config(cuda_config)
        .init()
        .await
        .unwrap();
    assert!(
        backend
            .llama_cpp()
            .unwrap()
            .server
            .device_config
            .gpu_count()
            == 2
    );
    let mut req = CompletionRequest::new(backend);
    req.prompt
        .add_user_message()
        .unwrap()
        .set_content("Hello, world!");

    let res = req.request().await.unwrap();
    println!("{res}");
}

#[tokio::test]
#[serial]
async fn test_cpu_only() {
    let backend = LlmInterface::llama_cpp().cpu_only().init().await.unwrap();
    assert!(
        backend
            .llama_cpp()
            .unwrap()
            .server
            .device_config
            .gpu_count()
            == 0
    );
    let mut req = CompletionRequest::new(backend);
    req.prompt
        .add_user_message()
        .unwrap()
        .set_content("Hello, world!");

    let res = req.request().await.unwrap();
    println!("{res}");
}

#[cfg(target_os = "macos")]
#[tokio::test]
#[serial]
async fn test_metal() {
    let metal_config = MetalConfig::new_from_ram_gb(5.0);
    let backend = LlmInterface::llama_cpp()
        .metal_config(metal_config)
        .init()
        .await
        .unwrap();
    assert!(
        backend
            .llama_cpp()
            .unwrap()
            .server
            .device_config
            .gpu_count()
            == 1
    );
    let mut req = CompletionRequest::new(backend);
    req.prompt
        .add_user_message()
        .unwrap()
        .set_content("Hello, world!");

    let res = req.request().await.unwrap();
    println!("{res}");
}