[Refactor][kv_offload] KV Offloading maintainability improvements (#40538)

Signed-off-by: Martin Hickey <martin.hickey@ie.ibm.com>
Co-authored-by: Or Ozeri <oro@il.ibm.com>
This commit is contained in:
Martin Hickey
2026-04-30 05:55:31 +03:00
committed by GitHub
co-authored by Or Ozeri
parent 0ff1bf9bb1
commit c42981d034
24 changed files with 446 additions and 440 deletions
@@ -18,15 +18,15 @@ from vllm.logger import init_logger
from vllm.utils.math_utils import cdiv
from vllm.v1.core.kv_cache_manager import KVCacheBlocks
from vllm.v1.core.sched.output import SchedulerOutput
from vllm.v1.kv_offload.abstract import (
from vllm.v1.kv_offload.base import (
GPULoadStoreSpec,
OffloadingManager,
OffloadingSpec,
OffloadKey,
ReqContext,
get_offload_block_hash,
make_offload_key,
)
from vllm.v1.kv_offload.mediums import GPULoadStoreSpec
from vllm.v1.kv_offload.spec import OffloadingSpec
from vllm.v1.outputs import KVConnectorOutput
from vllm.v1.request import Request
@@ -25,7 +25,7 @@ from vllm.v1.kv_cache_interface import (
MambaSpec,
UniformTypeKVCacheSpecs,
)
from vllm.v1.kv_offload.spec import (
from vllm.v1.kv_offload.base import (
CanonicalKVCacheRef,
CanonicalKVCaches,
CanonicalKVCacheTensor,