wehub-resource-sync
23 lines
658 B
ReStructuredText
23 lines
658 B
ReStructuredText
.. _recipes:
|
|
|
|
Recipes
|
|
=======
|
|
|
|
This section lists model architectures validated end-to-end with LMCache, with a
|
|
recipe page per architecture covering only the LMCache-specific configuration
|
|
that diverges from defaults. Engine-side documentation (how to serve the model
|
|
itself) lives with the serving engine; recipe pages link out rather than
|
|
duplicate.
|
|
|
|
For the generic LMCache + engine wiring (ports, remote hosts, sending a first
|
|
request), see :doc:`../getting_started/quickstart` -- recipes assume that page
|
|
as a prerequisite.
|
|
|
|
Recipes are grouped by attention architecture:
|
|
|
|
.. toctree::
|
|
:maxdepth: 1
|
|
|
|
uniform_attention_models
|
|
/mp/hybrid_models
|