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 14, 05:30 UTC
[Bugfix] Fix incorrect import of CacheConfig (#24631)
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
Cyrus Leung committed
Sep 11, 2025 at 16:48 UTC
6aeb1dab4a3585fb0a16f76b7aa2779fffdd5fc8
1 file changed
+1
-2
vllm/attention/layers/cross_attention.py
+1
-2
@@ -6,14 +6,13 @@ from typing import Optional
6
7
import numpy as np
8
import torch
9
-from transformers import CacheConfig
9
10
from vllm import envs
11
from vllm.attention.backends.abstract import (AttentionBackend,
12
AttentionMetadata, AttentionType)
13
from vllm.attention.layer import Attention
14
from vllm.attention.selector import get_attn_backend
16
-from vllm.config import VllmConfig
15
+from vllm.config import CacheConfig, VllmConfig
16
from vllm.logger import init_logger
17
from vllm.multimodal import MULTIMODAL_REGISTRY
18
from vllm.utils import cdiv