Feature Flags

PMetal uses Cargo feature flags to control which crates and capabilities are compiled.

Feature Matrix

Feature	Default	Crate	Description
core	Yes	pmetal-core	Foundation types, configs, traits
gguf	Yes	pmetal-gguf	GGUF format support
metal	Yes	pmetal-metal	Metal GPU kernels
hub	Yes	pmetal-hub	HuggingFace Hub integration
mlx	Yes	pmetal-mlx	MLX backend
models	Yes	pmetal-models	LLM architectures
lora	Yes	pmetal-lora	LoRA/QLoRA
trainer	Yes	pmetal-trainer	Training loops (pulls in data, distill)
easy	Yes	—	High-level builders (pulls in trainer, hub, data)
ane	Yes	—	Apple Neural Engine
data	Yes*	pmetal-data	Dataset loading (*via easy)
distill	Yes*	pmetal-distill	Knowledge distillation (*via trainer)
lora-metal-fused	No	—	~2× LoRA training speedup via fused Metal kernels
merge	No	pmetal-merge	Model merging strategies
vocoder	No	pmetal-vocoder	BigVGAN neural vocoder
distributed	No	pmetal-distributed	Distributed training
mhc	No	pmetal-mhc	Manifold-Constrained Hyper-Connections
serve	No	pmetal-serve	OpenAI-compatible inference server
full	No	—	All features

Usage

# Default features
cargo install pmetal

# With specific features
cargo install pmetal --features "merge,serve"

# All features
cargo install pmetal --features full

# Minimal build (no ANE)
cargo build --release --no-default-features --features dashboard

As a Library Dependency

[dependencies]
pmetal = { version = "0.1", features = ["easy"] }

# Or specific crates
pmetal-models = "0.1"
pmetal-trainer = "0.1"
pmetal-lora = "0.1"

See Also

• Installation — Build options
• Advanced SDK Usage — Crate-level API