API Reference emlx v#0.4.1

Low-level MLX NIF wrappers and the native Nx.Defn.Compiler for EMLX.Backend tensors.

OTP Application for EMLX.

EMLX backend for Nx tensors, providing GPU acceleration via Apple's MLX.

A handle to an emlx::Worker OS thread + mlx::core::Stream pair.

Topological traversal of Nx.Defn.Expr DAGs.

Single-kernel Metal shaders from mlx::fast, exposed as deftransform functions.

Elixir bindings for MLX array operations.

Shared utilities for the EMLX native compiler layer.

Fused native NIF wrappers for Qwen3 transformer inference.

Affine group-wise int2/int4/int8 quantization for Apple Silicon inference, plus microscaled floating-point modes ("mxfp4"/"mxfp8"/"nvfp4", via :mode on quantize/2 — see EMLX.quantize/2).

:telemetry events emitted by EMLX.