biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[Attention] Update attention imports (#29540)

Browse Source 
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
This commit is contained in:
Matthew Bonanni
2025-11-27 11:19:09 -05:00
committed by GitHub
parent cd007a53b4
commit fc1d8be3dc
38 changed files with 63 additions and 126 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
vllm/model_executor/layers/quantization/fp8.py
View File

@@ -14,6 +14,7 @@ import vllm.envs as envs
import vllm.model_executor.layers.fused_moe.modular_kernel as mk
from vllm import _custom_ops as ops
from vllm._aiter_ops import rocm_aiter_ops
from vllm.attention.layer import Attention
from vllm.distributed import get_tensor_model_parallel_world_size
from vllm.logger import init_logger
from vllm.model_executor.layers.batch_invariant import (
@@ -277,7 +278,6 @@ class Fp8Config(QuantizationConfig):
def get_xpu_quant_method(
self, layer: torch.nn.Module, prefix: str
) -> Optional["QuantizeMethodBase"]:
from vllm.attention.layer import Attention
from vllm.model_executor.layers.quantization.ipex_quant import (
XPUFp8LinearMethod,
XPUFp8MoEMethod,
@@ -307,8 +307,6 @@ class Fp8Config(QuantizationConfig):
def get_quant_method(
self, layer: torch.nn.Module, prefix: str
) -> Optional["QuantizeMethodBase"]:
from vllm.attention.layer import Attention # Avoid circular import
if current_platform.is_xpu():
return self.get_xpu_quant_method(layer, prefix)
if isinstance(layer, LinearBase):