Type Info#
Created On: Jun 06, 2025 | Last Updated On: Aug 14, 2025
The numerical properties of atorch.dtype can be accessed through either thetorch.finfo or thetorch.iinfo.
torch.finfo#
- classtorch.finfo#
Atorch.finfo is an object that represents the numerical properties of a floating pointtorch.dtype, (i.e.torch.float32,torch.float64,torch.float16, andtorch.bfloat16).This is similar tonumpy.finfo.
Atorch.finfo provides the following attributes:
Name | Type | Description |
|---|---|---|
bits | int | The number of bits occupied by the type. |
eps | float | The difference between 1.0 and the next smallest representable float larger than 1.0. |
max | float | The largest representable number. |
min | float | The smallest representable number (typically |
tiny | float | The smallest positive normal number. Equivalent to |
smallest_normal | float | The smallest positive normal number. See notes. |
resolution | float | The approximate decimal resolution of this type, i.e., |
Note
The constructor oftorch.finfo can be called without argument,in which case the class is created for the pytorch default dtype (as returned bytorch.get_default_dtype()).
Note
smallest_normal returns the smallestnormal number, but there are smallersubnormal numbers. See https://en.wikipedia.org/wiki/Denormal_numberfor more information.
torch.iinfo#
- classtorch.iinfo#
Atorch.iinfo is an object that represents the numerical properties of a integertorch.dtype (i.e.torch.uint8,torch.int8,torch.int16,torch.int32, andtorch.int64).This is similar tonumpy.iinfo.
Atorch.iinfo provides the following attributes:
Name | Type | Description |
|---|---|---|
bits | int | The number of bits occupied by the type. |
max | int | The largest representable number. |
min | int | The smallest representable number. |
