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 6, 19:15 UTC
[XPU] Support LoRA via torch.compile on XPU platform (#36962)
Signed-off-by: chzhang <chaojun.zhang@intel.com>
Chaojun Zhang committed
Mar 13, 2026 at 18:34 UTC
82f836d976f37657586a749372ea9fa432a62fce
1 file changed
+1
-3
vllm/platforms/xpu.py
+1
-3
@@ -167,7 +167,7 @@ class XPUPlatform(Platform):
167
cache_config.block_size = 64
168
169
# lazy import to avoid circular import
170
- from vllm.config import CompilationMode, CUDAGraphMode
170
+ from vllm.config import CUDAGraphMode
171
172
compilation_config = vllm_config.compilation_config
173
if compilation_config.compile_sizes is None:
@@ -200,8 +200,6 @@ class XPUPlatform(Platform):
200
"falling back to PIECEWISE graph mode on XPU platform."
201
)
202
203
- if vllm_config.lora_config is not None:
204
- compilation_config.mode = CompilationMode.NONE
203
# check and update parallel config
204
parallel_config = vllm_config.parallel_config
205
# Only override worker_cls if it's still the default "auto"