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 11, 03:45 UTC
[Misc] Fix grammar errors in comments and messages (#31115)
Signed-off-by: c0de128 <kevin.mckay@outlook.com>
Kevin McKay committed
Dec 21, 2025 at 23:14 UTC
42b42824ae8223fa84e7545709572c1fd231436b
2 files changed
+4
-4
tests/quantization/test_compressed_tensors.py
+3
-3
@@ -83,7 +83,7 @@ def test_compressed_tensors_w8a8_static_setup(vllm_runner, model_args):
83
current_platform.is_rocm()
84
and model_path not in ROCM_TRITON_SCALED_MM_SUPPORTED_INT8_MODEL
85
):
86
- pytest.skip(f"Skip model {model_path} as it is not support on ROCm.")
86
+ pytest.skip(f"Skip model {model_path} as it is not supported on ROCm.")
87
88
with vllm_runner(model_path, enforce_eager=True) as llm:
89
@@ -161,7 +161,7 @@ def test_compressed_tensors_w8a8_logprobs(
161
current_platform.is_rocm()
162
and model_path not in ROCM_TRITON_SCALED_MM_SUPPORTED_INT8_MODEL
163
):
164
- pytest.skip(f"Skip model {model_path} as it is not support on ROCm.")
164
+ pytest.skip(f"Skip model {model_path} as it is not supported on ROCm.")
165
166
if use_aiter:
167
if model_path not in ROCM_AITER_SUPPORTED_INT8_MODEL:
@@ -231,7 +231,7 @@ def test_compressed_tensors_w8a8_dynamic_per_token(
231
current_platform.is_rocm()
232
and model_path not in ROCM_TRITON_SCALED_MM_SUPPORTED_INT8_MODEL
233
):
234
- pytest.skip(f"Skip model {model_path} as it is not support on ROCm.")
234
+ pytest.skip(f"Skip model {model_path} as it is not supported on ROCm.")
235
236
if use_aiter:
237
if model_path not in ROCM_AITER_SUPPORTED_INT8_MODEL:
vllm/attention/ops/merge_attn_states.py
+1
-1
@@ -15,7 +15,7 @@ def merge_attn_states(
15
output_lse: torch.Tensor | None = None,
16
) -> None:
17
# NOTE(DefTruth): Currently, custom merge_attn_states CUDA kernel
18
- # is not support for FP8 dtype, fallback to use Triton kernel.
18
+ # does not support FP8 dtype, fallback to use Triton kernel.
19
def supported_dtypes(o: torch.Tensor) -> bool:
20
return o.dtype in [torch.float32, torch.half, torch.bfloat16]
21