aether-llm 0.7.0

Multi-provider LLM abstraction layer for the Aether AI agent framework
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

#![doc = include_str!(concat!(env!("OUT_DIR"), "/docs/llamacpp.md"))]

use super::util::get_local_config;
use crate::providers::openai::OpenAiChatProvider;
use crate::{ProviderConnectionConfig, ProviderFactory, Result};
use async_openai::{Client, config::OpenAIConfig};

pub struct LlamaCppProvider {
    client: Client<OpenAIConfig>,
}

impl LlamaCppProvider {
    pub fn new(base_url: &str) -> Self {
        Self { client: Client::with_config(get_local_config(base_url)) }
    }
}

impl Default for LlamaCppProvider {
    fn default() -> Self {
        Self { client: Client::with_config(get_local_config("http://localhost:8080/v1")) }
    }
}

impl ProviderFactory for LlamaCppProvider {
    async fn from_env() -> Result<Self> {
        Self::from_env_with_connection(ProviderConnectionConfig::default()).await
    }

    async fn from_env_with_connection(connection: ProviderConnectionConfig) -> Result<Self> {
        let base_url = connection.base_url.as_deref().unwrap_or("http://localhost:8080/v1");
        Ok(Self { client: Client::with_config(get_local_config(base_url)) })
    }

    fn with_model(self, _model: &str) -> Self {
        // LlamaCpp doesn't support model selection - it serves a single model
        self
    }
}

impl OpenAiChatProvider for LlamaCppProvider {
    type Config = OpenAIConfig;

    fn client(&self) -> &Client<Self::Config> {
        &self.client
    }

    fn model(&self) -> &'static str {
        "" // llama.cpp server serves a single model on boot and does not allow swapping models
    }

    fn provider_name(&self) -> &'static str {
        "LlamaCpp"
    }
}