GPU Model Runner V2 (#25266)
Signed-off-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>
This commit is contained in:
@@ -231,6 +231,7 @@ if TYPE_CHECKING:
|
||||
VLLM_DISABLE_SHARED_EXPERTS_STREAM: bool = False
|
||||
VLLM_SHARED_EXPERTS_STREAM_TOKEN_THRESHOLD: int = 256
|
||||
VLLM_COMPILE_CACHE_SAVE_FORMAT: Literal["binary", "unpacked"] = "binary"
|
||||
VLLM_USE_V2_MODEL_RUNNER: bool = False
|
||||
|
||||
|
||||
def get_default_cache_root():
|
||||
@@ -1522,6 +1523,10 @@ environment_variables: dict[str, Callable[[], Any]] = {
|
||||
"VLLM_COMPILE_CACHE_SAVE_FORMAT": env_with_choices(
|
||||
"VLLM_COMPILE_CACHE_SAVE_FORMAT", "binary", ["binary", "unpacked"]
|
||||
),
|
||||
# Flag to enable v2 model runner.
|
||||
"VLLM_USE_V2_MODEL_RUNNER": lambda: bool(
|
||||
int(os.getenv("VLLM_USE_V2_MODEL_RUNNER", "0"))
|
||||
),
|
||||
}
|
||||
|
||||
# --8<-- [end:env-vars-definition]
|
||||
|
||||
Reference in New Issue
Block a user