[Deprecation] Remove deprecated task, seed and MM settings (#30397)

Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
This commit is contained in:
Cyrus Leung
2025-12-10 19:59:39 -08:00
committed by GitHub
parent 5a87d8b9b1
commit 7e24e5d4d6
16 changed files with 25 additions and 276 deletions
-5
View File
@@ -72,7 +72,6 @@ if TYPE_CHECKING:
VLLM_MAX_AUDIO_CLIP_FILESIZE_MB: int = 25
VLLM_VIDEO_LOADER_BACKEND: str = "opencv"
VLLM_MEDIA_CONNECTOR: str = "http"
VLLM_MM_INPUT_CACHE_GIB: int = 4
VLLM_TARGET_DEVICE: str = "cuda"
VLLM_MAIN_CUDA_VERSION: str = "12.9"
VLLM_FLOAT32_MATMUL_PRECISION: Literal["highest", "high", "medium"] = "highest"
@@ -786,9 +785,6 @@ environment_variables: dict[str, Callable[[], Any]] = {
# imported at runtime.
# If a non-existing backend is used, an AssertionError will be thrown.
"VLLM_MEDIA_CONNECTOR": lambda: os.getenv("VLLM_MEDIA_CONNECTOR", "http"),
# [DEPRECATED] Cache size (in GiB per process) for multimodal input cache
# Default is 4 GiB per API process + 4 GiB per engine core process
"VLLM_MM_INPUT_CACHE_GIB": lambda: int(os.getenv("VLLM_MM_INPUT_CACHE_GIB", "4")),
# Path to the XLA persistent cache directory.
# Only used for XLA devices such as TPUs.
"VLLM_XLA_CACHE_PATH": lambda: os.path.expanduser(
@@ -1681,7 +1677,6 @@ def compile_factors() -> dict[str, object]:
"VLLM_MEDIA_CONNECTOR",
"VLLM_ASSETS_CACHE",
"VLLM_ASSETS_CACHE_MODEL_CLEAN",
"VLLM_MM_INPUT_CACHE_GIB",
"VLLM_WORKER_MULTIPROC_METHOD",
"VLLM_ENABLE_V1_MULTIPROCESSING",
"VLLM_V1_OUTPUT_PROC_CHUNK_SIZE",