Skip to content

vllm_omni.diffusion.diffusion_kv.config ¶

DiffusionKVCacheMode ¶

Bases: str, Enum

Migration mode for diffusion KV ownership.

DENSE_LEGACY `class-attribute` `instance-attribute` ¶

DENSE_LEGACY = 'dense_legacy'

PAGED_SCHEDULER `class-attribute` `instance-attribute` ¶

PAGED_SCHEDULER = 'paged_scheduler'

PAGED_WORKER_LOCAL `class-attribute` `instance-attribute` ¶

PAGED_WORKER_LOCAL = 'paged_worker_local'

is_scheduler_paged_kv_mode ¶

is_scheduler_paged_kv_mode(
    mode: DiffusionKVCacheMode,
) -> bool

Return whether a parsed cache mode uses Scheduler-owned paging.

parse_diffusion_kv_cache_mode ¶

parse_diffusion_kv_cache_mode(
    value: object,
) -> DiffusionKVCacheMode

Parse a selectable Diffusion KV ownership mode.

PAGED_WORKER_LOCAL remains in the enum to document the migration topology, but it is not a valid production selection until that runtime is wired. Keep this validation shared by structured and runtime configs so an unsupported mode fails before managed-process startup.