We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
There was an error while loading. Please reload this page.
1 parent 8008fd3 commit f91ead3Copy full SHA for f91ead3
vllm/platforms/xpu.py
@@ -7,7 +7,6 @@
7
import torch
8
9
import vllm.envs as envs
10
-from vllm.config import CUDAGraphMode
11
from vllm.logger import init_logger
12
from vllm.utils import DEFAULT_MAX_NUM_BATCHED_TOKENS
13
@@ -109,6 +108,8 @@ def check_and_update_config(cls, vllm_config: VllmConfig) -> None:
109
108
and not cls.device_support_bf16():
110
model_config.dtype = torch.float16
111
+ # lazy import to avoid circular import
112
+ from vllm.config import CUDAGraphMode
113
compilation_config = vllm_config.compilation_config
114
if compilation_config.cudagraph_mode is None or \
115
compilation_config.cudagraph_mode.max_cudagraph_mode() \
0 commit comments