[Bugfix] Register VLLM_BATCH_INVARIANT in envs.py to fix spurious unknown env var warning (#35007)
Signed-off-by: Ranran <1012869439@qq.com> Signed-off-by: Ranran <hzz5361@psu.edu> Signed-off-by: ran <hzz5361@psu.edu> Co-authored-by: Wentao Ye <44945378+yewentao256@users.noreply.github.com>
This commit is contained in:
@@ -6,6 +6,7 @@ from typing import Any
|
||||
|
||||
import torch
|
||||
|
||||
import vllm.envs as envs
|
||||
from vllm.logger import init_logger
|
||||
from vllm.platforms import current_platform
|
||||
from vllm.triton_utils import tl, triton
|
||||
@@ -986,21 +987,6 @@ def enable_batch_invariant_mode():
|
||||
torch.backends.cuda.preferred_blas_library(backend="cublaslt")
|
||||
|
||||
|
||||
def _read_vllm_batch_invariant() -> bool:
|
||||
val = os.getenv("VLLM_BATCH_INVARIANT", "0")
|
||||
try:
|
||||
return int(val) != 0
|
||||
except ValueError:
|
||||
return False
|
||||
|
||||
|
||||
VLLM_BATCH_INVARIANT: bool = _read_vllm_batch_invariant()
|
||||
|
||||
|
||||
def vllm_is_batch_invariant() -> bool:
|
||||
return VLLM_BATCH_INVARIANT
|
||||
|
||||
|
||||
def override_envs_for_invariance(
|
||||
attention_backend: AttentionBackendEnum | None,
|
||||
):
|
||||
@@ -1059,7 +1045,7 @@ def init_batch_invariance(
|
||||
attention_backend: AttentionBackendEnum | None,
|
||||
):
|
||||
# this will hit all the csrc overrides as well
|
||||
if vllm_is_batch_invariant():
|
||||
if envs.VLLM_BATCH_INVARIANT:
|
||||
override_envs_for_invariance(attention_backend)
|
||||
enable_batch_invariant_mode()
|
||||
|
||||
|
||||
Reference in New Issue
Block a user