[Attention] Update attention imports (#29540)
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
This commit is contained in:
@@ -8,6 +8,7 @@ import numpy as np
|
||||
import torch
|
||||
import torch.nn as nn
|
||||
|
||||
from vllm.attention.backends.registry import AttentionBackendEnum
|
||||
from vllm.config import (
|
||||
CompilationMode,
|
||||
CUDAGraphMode,
|
||||
@@ -157,8 +158,6 @@ class EagleProposer:
|
||||
)
|
||||
|
||||
# Determine allowed attention backends once during initialization.
|
||||
from vllm.attention.backends.registry import AttentionBackendEnum
|
||||
|
||||
self.allowed_attn_types: tuple | None = None
|
||||
if current_platform.is_rocm():
|
||||
rocm_types = [TritonAttentionMetadata, FlashAttentionMetadata]
|
||||
|
||||
Reference in New Issue
Block a user