[Misc] Remove redundant attention var constants (#29650)
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
This commit is contained in:
@@ -11,7 +11,6 @@ import pytest
|
||||
from tests.quantization.utils import is_quant_method_supported
|
||||
from vllm.attention.utils.fa_utils import flash_attn_supports_fp8
|
||||
from vllm.platforms import current_platform
|
||||
from vllm.utils import STR_BACKEND_ENV_VAR
|
||||
from ..utils import check_logprobs_close
|
||||
|
||||
|
||||
@@ -76,7 +75,7 @@ def test_models(
|
||||
|
||||
with monkeypatch.context() as m:
|
||||
m.setenv("TOKENIZERS_PARALLELISM", "true")
|
||||
m.setenv(STR_BACKEND_ENV_VAR, backend)
|
||||
m.setenv("VLLM_ATTENTION_BACKEND", backend)
|
||||
|
||||
MAX_MODEL_LEN = 1024
|
||||
NUM_LOG_PROBS = 8
|
||||
|
||||
Reference in New Issue
Block a user