Skip to main content

Ctrl+K

Install

Build and Install

Usage

Quick Start Guide
Lazy Evaluation
Unified Memory
Indexing Arrays
Writing a Fast KV Cache
Saving and Loading Arrays
Function Transforms
Compilation
Conversion to NumPy and Other Frameworks
Numerical Precision
Environment Variables
Distributed Communication
Using Streams
Exporting Functions

Examples

Linear Regression
Multi-Layer Perceptron
LLM inference
Data Parallelism
Tensor Parallelism

Python API Reference

Array
Data Types
Devices and Streams
Export Functions
Operations
Random
Transforms
Fast
FFT
Linear Algebra
Metal
CUDA
- mlx.core.cuda.is_available
Memory Management
Neural Networks
Optimizers
Distributed Communication
Tree Utils
Print Options

C++ API Reference

Operations

Further Reading

Custom Extensions in MLX
Metal Debugger
Metal Logging
Custom Metal Kernels
Using MLX in C++

.rst

mlx.core.default_device

Contents

default_device()

mlx.core.default_device#

default_device() → Device#: Get the default device.

previous

mlx.core.ThreadLocalStream

next

mlx.core.set_default_device

Contents

default_device()

By MLX Contributors

© Copyright 2023, Apple.