Module scheme 

Types representing the quantization scheme

Structs

BlockScale

The per-block scale level of a QuantScheme: one scale per block of values.

BlockSize

Copyable block size, specialized version of SmallVec.

F32Grid

A narrower float format’s grid, laid over the f32 bit pattern.

QuantScheme

Describes a quantization scheme/configuration.

SubnormalRange

Where a format’s subnormals begin and how far apart they are, in f32.

Enums

QuantMode

Strategy used to quantize values.

QuantStore

Data type used to stored quantized values.

QuantValue

Data type used to represent quantized values.

ScaleDtype

The data type a scale level stores its scales in.

Functions

f32_to_ue8m0

A ue8m0 scale as its stored byte: the code is the exponent, biased by 127.

ue8m0_to_f32

The value a ue8m0 byte stands for. Inverse of f32_to_ue8m0 on every code it produces.