car-inference 0.56.1

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
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40

// The local-backend abstraction (trait + dispatch) is platform-neutral; only
// its implementors are cfg-gated, exactly as `ProtocolHandler` is unconditional
// while the per-protocol wiring varies.
pub mod local;

#[cfg(not(all(target_os = "macos", target_arch = "aarch64", not(car_skip_mlx))))]
pub mod candle;
#[cfg(not(all(target_os = "macos", target_arch = "aarch64", not(car_skip_mlx))))]
pub mod embedding;
#[cfg(not(all(target_os = "macos", target_arch = "aarch64", not(car_skip_mlx))))]
pub mod moe;

// Subscription-backed Codex inference is a platform-neutral subprocess
// backend. It stays separate from the agentic `car-external-agents` adapter.
pub mod codex_cli;

// FoundationModels is the only module here that's also on iOS —
// every `mlx*` / `external_*` sibling stays macOS-aarch64 because
// `mlx-rs` itself can't compile for iOS.
#[cfg(all(target_os = "macos", target_arch = "aarch64", not(car_skip_mlx)))]
pub mod external_flux;
#[cfg(all(target_os = "macos", target_arch = "aarch64", not(car_skip_mlx)))]
pub mod external_ltx;
#[cfg(any(
    all(target_os = "macos", target_arch = "aarch64", not(car_skip_mlx)),
    all(target_os = "ios", target_arch = "aarch64")
))]
pub mod foundation_models;
#[cfg(all(target_os = "macos", target_arch = "aarch64", not(car_skip_mlx)))]
pub mod mlx_vlm_cli;
#[cfg(all(target_os = "macos", target_arch = "aarch64", not(car_skip_mlx)))]
pub mod swift_lm;

#[cfg(not(all(target_os = "macos", target_arch = "aarch64", not(car_skip_mlx))))]
pub use self::candle::CandleBackend;
#[cfg(not(all(target_os = "macos", target_arch = "aarch64", not(car_skip_mlx))))]
pub use self::embedding::EmbeddingBackend;

#[cfg(all(target_os = "macos", target_arch = "aarch64", not(car_skip_mlx)))]
pub use self::swift_lm::SwiftLmBackend;