mlx-bun library API
Documentation
Preparing search index...
@mlx-bun/inference
@mlx-bun/inference/state
activeQuantViews
Function activeQuantViews
activeQuantViews
(
keys
:
QuantizedTensor
,
values
:
QuantizedTensor
,
length
:
number
,
step
:
DecodeStepInputs
,
)
:
MlxArray
[]
The six active-prefix planes (keys then values) a concat fetch feeds a compiled decode step.
Parameters
keys
:
QuantizedTensor
values
:
QuantizedTensor
length
:
number
step
:
DecodeStepInputs
Returns
MlxArray
[]
Settings
Member Visibility
Protected
Inherited
External
Theme
OS
Light
Dark
Documentation
mlx-bun library API
Loading...
The six active-prefix planes (keys then values) a concat fetch feeds a compiled decode step.