A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 6, 19:15 UTC
[Bugfix] fix main branch pre-commit error (1 line change) (#36897)
Signed-off-by: SoluMilken <ypiheyn.imm02g@g2.nctu.edu.tw>
SoluMilken committed
Mar 13, 2026 at 00:08 UTC
85199f9681af6656c3f61f982e826f61664eb2af
1 file changed
+1
-1
tests/kernels/test_fused_recurrent_packed_decode.py
+1
-1
@@ -10,7 +10,7 @@ from vllm.model_executor.layers.fla.ops import (
10
)
11
12
13
-@pytest.mark.skipif(torch.cuda.device_count() < 1, reason="Need CUDA device")
13
+@pytest.mark.skipif(not torch.cuda.is_available(), reason="Need CUDA device")
14
@pytest.mark.parametrize("dtype", [torch.float16, torch.bfloat16, torch.float32])
15
@pytest.mark.parametrize("strided_mixed_qkv", [False, True])
16
def test_fused_recurrent_packed_decode_matches_reference(