mlx-bun library API
Documentation
Preparing search index...
@mlx-bun/inference
@mlx-bun/inference/state/kv-scheme
kvQuantBytesPerElement
Function kvQuantBytesPerElement
kvQuantBytesPerElement
(
bits
:
number
,
groupSize
:
number
)
:
number
Bytes per stored affine-quantized KV element, including scale and bias.
Parameters
bits
:
number
groupSize
:
number
Returns
number
Settings
Member Visibility
Protected
Inherited
External
Theme
OS
Light
Dark
Documentation
mlx-bun library API
Loading...
Bytes per stored affine-quantized KV element, including scale and bias.