[package]
name = "ferrum-cli"
version.workspace = true
edition.workspace = true
authors.workspace = true
license.workspace = true
description = "CLI for Ferrum — a Rust-native LLM inference engine"
readme = "../../README.md"
[[bin]]
name = "ferrum"
path = "src/main.rs"
[dependencies]
ferrum-types = { workspace = true }
ferrum-interfaces = { workspace = true }
ferrum-engine = { workspace = true }
ferrum-models = { workspace = true }
ferrum-runtime = { workspace = true }
ferrum-scheduler = { workspace = true }
ferrum-server = { workspace = true }
candle-core = { workspace = true }
tokenizers = { workspace = true }
clap = { version = "4.4", features = ["derive", "color", "env"] }
tokio = { workspace = true, features = ["full"] }
async-trait = { workspace = true }
serde = { workspace = true }
serde_json = { workspace = true }
serde_yaml = "0.9"
anyhow = { workspace = true }
thiserror = { workspace = true }
tracing = { workspace = true }
tracing-subscriber = { version = "0.3", features = ["env-filter", "json"] }
config = { workspace = true }
toml = { workspace = true }
walkdir = "2.4"
console = "0.15"
indicatif = "0.17"
dialoguer = "0.11"
colored = "2.1"
chrono = { workspace = true }
uuid = { workspace = true }
reqwest = { version = "0.12", default-features = false, features = ["json", "stream", "rustls-tls"] }
tokio-stream = "0.1"
futures = "0.3"
criterion = { version = "0.5", features = ["html_reports"] }
num_cpus = { workspace = true }
dirs = { workspace = true }
shellexpand = "3.1"
[features]
default = []
benchmark = []
dev-tools = []
interactive = []
metal = ["ferrum-engine/metal"]
cuda = ["ferrum-engine/cuda"]
marlin = ["cuda"]
tensor-parallel = ["cuda"]