Python module

dtype

Provides data type definitions for tensors in MAX Engine. These data types are essential for defining the precision and memory layout of tensor data when working with machine learning models.

This module defines the DType enum, which represents all supported tensor data types in MAX Engine, including:

Integer types (signed and unsigned): int8 | uint8 | int16 | uint16 | int32 | uint32 | int64 | uint64
Floating-point types: float8 variants | float16 | bfloat16 | float32 | float64
Boolean type

The module also provides utilities for converting between MAX Engine data types and NumPy dtypes, making it easy to interoperate with the NumPy ecosystem.

import numpy as np
from max.dtype import DType

tensor = np.zeros((2, 3), dtype=DType.float32.to_numpy())

# Convert NumPy dtype to MAX DType
array = np.ones((4, 4), dtype=np.float16)
max_dtype = DType.from_numpy(array.dtype)

# Check properties of data types
is_float = DType.float32.is_float()  # True
is_int = DType.int64.is_integral()   # True
size = DType.float64.size_in_bytes   # 8
import numpy as np
from max.dtype import DType

tensor = np.zeros((2, 3), dtype=DType.float32.to_numpy())

# Convert NumPy dtype to MAX DType
array = np.ones((4, 4), dtype=np.float16)
max_dtype = DType.from_numpy(array.dtype)

# Check properties of data types
is_float = DType.float32.is_float()  # True
is_int = DType.int64.is_integral()   # True
size = DType.float64.size_in_bytes   # 8

`DType`

class max.dtype.DType(value, names=<not given>, *values, module=None, qualname=None, type=None, start=1, boundary=None)

The tensor data type.

`align`

property align*: int*

Returns the alignment of the dtype.

`bfloat16`

bfloat16 = 71

`bool`

bool = 1

`float16`

float16 = 70

`float32`

float32 = 72

`float64`

float64 = 73

`float8_e4m3`

float8_e4m3 = 65

`float8_e4m3fn`

float8_e4m3fn = 66

`float8_e4m3fnuz`

float8_e4m3fnuz = 67

`float8_e5m2`

float8_e5m2 = 68

`float8_e5m2fnuz`

float8_e5m2fnuz = 69

`from_numpy()`

from_numpy() → DType

Converts a NumPy dtype to the corresponding DType.

Parameters:

dtype (np.dtype) – The NumPy dtype to convert.
Returns:

The corresponding DType enum value.
Return type:

DType
Raises:

ValueError – If the input dtype is not supported.

`int16`

int16 = 137

`int32`

int32 = 139

`int64`

int64 = 141

`int8`

int8 = 135

`is_float`

is_float

Returns true if the dtype is floating point.

`is_float8`

is_float8

Returns true if the dtype is any variant of float8.

`is_half`

is_half

Returns true if the dtype is half-precision floating point.

`is_integral`

is_integral

Returns true if the dtype is an integer.

`size_in_bytes`

property size_in_bytes

Returns the size of the dtype in bytes.

`to_numpy()`

to_numpy() → dtype

Converts this DType to the corresponding NumPy dtype.

Returns:

The corresponding NumPy dtype object.
Return type:

DType
Raises:

ValueError – If the dtype is not supported.

`uint16`

uint16 = 136

`uint32`

uint32 = 138

`uint64`

uint64 = 140

`uint8`

uint8 = 134

`max_to_torch_type()`

max.dtype.max_to_torch_type(dtype: DType) → dtype

Converts a MAX dtype.DType to a torch.dtype.

Parameters:

dtype (dtype.DType) – MAX dtype.DType.
Returns:

The corresponding torch.dtype.
Return type:

torch.dtype

`torch_to_max_type()`

max.dtype.torch_to_max_type(dtype: dtype) → DType

Converts a torch.dtype to a MAX dtype.DType.

Parameters:

dtype (torch.dtype) – PyTorch torch.dtype.
Returns:

The corresponding dtype.DType.
Return type:

dtype.DType

DType​

align​

bfloat16​

bool​

float16​

float32​

float64​

float8_e4m3​

float8_e4m3fn​

float8_e4m3fnuz​

float8_e5m2​

float8_e5m2fnuz​

from_numpy()​

int16​

int32​

int64​

int8​

is_float​

is_float8​

is_half​

is_integral​

size_in_bytes​

to_numpy()​

uint16​

uint32​

uint64​

uint8​

max_to_torch_type()​

torch_to_max_type()​