car-inference 0.15.0

Local model inference for CAR — Candle backend with Qwen3 models
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

//! Embedding types for semantic similarity via Qwen3-Embedding.
//!
//! The actual embedding logic lives in `backend::EmbeddingBackend`.
//! This module defines the request type used by `InferenceEngine::embed()`.

use serde::{Deserialize, Serialize};

/// An embedding request.
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct EmbedRequest {
    /// Texts to embed.
    pub texts: Vec<String>,
    /// Reserved for API compatibility. The embedding model is always used.
    #[serde(default)]
    pub model: Option<String>,
    /// Optional instruction for query-side embedding (Qwen3-Embedding format).
    /// Default: "Retrieve relevant memory facts"
    #[serde(default)]
    pub instruction: Option<String>,
    /// Whether these texts are queries (true) or documents (false).
    /// Queries get the Instruct:/Query: prefix; documents are embedded raw.
    #[serde(default)]
    pub is_query: bool,
}