Skip to content

GitLab

Unverified Commit b72af8f1 authored by zhaoyang-star's avatar zhaoyang-star Committed by GitHub
Browse files

Fix error when tp > 1 (#2644) 


Co-authored-by: default avatarzhaoyang-star <zhao.yang16@zte.com.cn>
parent 9090bf02
Hide whitespace changes
Inline Side-by-side
Showing with 2 additions and 3 deletions
vllm/engine/llm_engine.py
View file @ b72af8f1
...@@ -236,7 +236,6 @@ class LLMEngine: ...@@ -236,7 +236,6 @@ class LLMEngine:
model_config = copy.deepcopy(self.model_config) model_config = copy.deepcopy(self.model_config)
parallel_config = copy.deepcopy(self.parallel_config) parallel_config = copy.deepcopy(self.parallel_config)
scheduler_config = copy.deepcopy(self.scheduler_config) scheduler_config = copy.deepcopy(self.scheduler_config)
cache_config = copy.deepcopy(self.cache_config)
for rank, (worker, (node_id, for rank, (worker, (node_id,
_)) in enumerate(zip(self.workers, _)) in enumerate(zip(self.workers,
...@@ -252,7 +251,7 @@ class LLMEngine: ...@@ -252,7 +251,7 @@ class LLMEngine:
rank, rank,
distributed_init_method, distributed_init_method,
lora_config=self.lora_config, lora_config=self.lora_config,
cache_config=cache_config, kv_cache_dtype=self.cache_config.cache_dtype,
)) ))
driver_rank = 0 driver_rank = 0
...@@ -265,7 +264,7 @@ class LLMEngine: ...@@ -265,7 +264,7 @@ class LLMEngine:
driver_rank, driver_rank,
distributed_init_method, distributed_init_method,
lora_config=self.lora_config, lora_config=self.lora_config,
cache_config=cache_config, kv_cache_dtype=self.cache_config.cache_dtype,
is_driver_worker=True, is_driver_worker=True,
) )
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment