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 11, 05:00 UTC
[BugFix] Spec decode with VLLM_ENABLE_V1_MULTIPROCESSING=0 (#30319)
Signed-off-by: Chen Zhang <zhangch99@outlook.com>
Chen Zhang committed
Dec 18, 2025 at 11:47 UTC
24b65eff0da0c8c4422f9cff6bf35f80a11c0274
1 file changed
+2
-1
vllm/v1/engine/core_client.py
+2
-1
@@ -268,7 +268,8 @@ class InprocClient(EngineCoreClient):
268
self.engine_core = EngineCore(*args, **kwargs)
269
270
def get_output(self) -> EngineCoreOutputs:
271
- outputs, _ = self.engine_core.step_fn()
271
+ outputs, model_executed = self.engine_core.step_fn()
272
+ self.engine_core.post_step(model_executed=model_executed)
273
return outputs and outputs.get(0) or EngineCoreOutputs()
274
275
def get_supported_tasks(self) -> tuple[SupportedTask, ...]: