MLX/MLXArray

An N dimensional array. The main type in mlx.

Introduction

MLXArray is an N dimension array that can contain a variety of data types (DType). MLXArray supports a wide range of functions and operations to manipulate it and is used as the basis for the MLXNN module which implements many layers that are the basis for building machine learning models.

// create an array from a swift array
let a1 = MLXArray([1, 2, 3])

// this holds Int32
print(a1.dtype)

// and has a shape of [3]
print(a1.shape)

// there are a variety of operators and functions that can be used
let a2 = sqrt(a1 * 3)

Thread Safety

MLXArray is not thread safe.

Although MLXArray looks like a normal multidimensional array, it is actually far more sophisticated. It actually holds a promise for future computations, see doc:lazy-evaluation and is thus not thread safe. For example:

let a: MLXArray
let b: MLXArray

let c = a + b

c is not the result of a + b but rather a graph representing a and b (which in turn may be large unresolved graphs) and the addition operator that combines them. It is not safe to create c in one thread and consume/evaluate it in another.

Memory Safety

MLXArray is not memory safe.

Unlike swift Array, MLXArray is not memory safe -- use caution when using indexing operators.

For example:

let a = MLXArray([0])
print(a[10000])

Will print a random value from outside a.

Topics

Creation

• doc:initialization

Arithmetic Operators and Functions

• doc:arithmetic
• doc:convolution
• doc:cumulative

Indexing

• doc:indexes
• doc:indexing
• doc:sorting

Logical Operations

• doc:logical

Shapes

Some methods allow you to manipulate the shape of the array. These methods change the size and shape of the dimensions without changing the number of elements or contents of the array.

• doc:shapes
• doc:reduction

Conversion and Data Types

• doc:conversion