[package]
edition = "2024"
name = "llama-runner"
version = "2.3.0"
build = false
autolib = false
autobins = false
autoexamples = false
autotests = false
autobenches = false
description = "A straightforward Rust library for running llama.cpp models locally on device"
homepage = "https://github.com/zhufucdev/llama-runner"
readme = "README.md"
keywords = [
"llama",
"language-model",
"llm",
"ai",
]
categories = [
"text-processing",
"science",
]
license = "Apache-2.0"
repository = "https://github.com/zhufucdev/llama-runner"
[features]
cuda = ["llama-cpp-2/cuda"]
mcp = [
"dep:minijinja",
"dep:minijinja-contrib",
"dep:rmcp",
]
metal = ["llama-cpp-2/metal"]
rocm = ["llama-cpp-2/rocm"]
vulkan = ["llama-cpp-2/vulkan"]
[lib]
name = "llama_runner"
path = "src/lib.rs"
[dependencies.encoding_rs]
version = "0.8"
[dependencies.hf-hub]
version = "0.5"
features = ["tokio"]
[dependencies.image]
version = "0.25"
[dependencies.llama-cpp-2]
version = "0.1.143"
features = [
"mtmd",
"llguidance",
]
[dependencies.log]
version = "0.4"
[dependencies.minijinja]
version = "2.19"
features = ["json"]
optional = true
[dependencies.minijinja-contrib]
version = "2.19"
features = ["pycompat"]
optional = true
[dependencies.rand]
version = "0.10"
[dependencies.rmcp]
version = "1.3"
features = ["server"]
optional = true
default-features = false
[dependencies.serde]
version = "1.0"
features = ["derive"]
[dependencies.serde_json]
version = "1.0"
[dependencies.strum]
version = "0.28"
features = ["derive"]
[dependencies.thiserror]
version = "2.0"
[dependencies.tokio]
version = "1.50"