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 17, 06:45 UTC
[BugFix] Set engine_use_ray=True when TP>1 (#1531)
ljss committed
Nov 1, 2023 at 17:14 UTC
5687d584fe4825d59a6fd9ecc5840e1df04e380f
1 file changed
+1
-1
vllm/engine/async_llm_engine.py
+1
-1
@@ -484,7 +484,7 @@ class AsyncLLMEngine:
484
distributed_init_method, placement_group = initialize_cluster(
485
parallel_config, engine_args.engine_use_ray)
486
# Create the async LLM engine.
487
- engine = cls(engine_args.worker_use_ray,
487
+ engine = cls(parallel_config.worker_use_ray,
488
engine_args.engine_use_ray,
489
*engine_configs,
490
distributed_init_method,