From 82f836d976f37657586a749372ea9fa432a62fce Mon Sep 17 00:00:00 2001 From: Chaojun Zhang Date: Fri, 13 Mar 2026 18:34:59 +0800 Subject: [PATCH] [XPU] Support LoRA via torch.compile on XPU platform (#36962) Signed-off-by: chzhang --- vllm/platforms/xpu.py | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) diff --git a/vllm/platforms/xpu.py b/vllm/platforms/xpu.py index b7bcee4dd..5d39dfceb 100644 --- a/vllm/platforms/xpu.py +++ b/vllm/platforms/xpu.py @@ -167,7 +167,7 @@ class XPUPlatform(Platform): cache_config.block_size = 64 # lazy import to avoid circular import - from vllm.config import CompilationMode, CUDAGraphMode + from vllm.config import CUDAGraphMode compilation_config = vllm_config.compilation_config if compilation_config.compile_sizes is None: @@ -200,8 +200,6 @@ class XPUPlatform(Platform): "falling back to PIECEWISE graph mode on XPU platform." ) - if vllm_config.lora_config is not None: - compilation_config.mode = CompilationMode.NONE # check and update parallel config parallel_config = vllm_config.parallel_config # Only override worker_cls if it's still the default "auto"