biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[2/N] executor pass the complete config to worker/modelrunner (#9938)

Browse Source 
Signed-off-by: youkaichao <youkaichao@gmail.com>
Co-authored-by: Nick Hill <nhill@redhat.com>
This commit is contained in:
youkaichao
2024-11-02 07:35:05 -07:00
committed by GitHub
parent 1d4cfe2be1
commit e893795443
44 changed files with 249 additions and 579 deletions
Download Patch File Download Diff File Expand all files Collapse all files

11
vllm/v1/executor/gpu_executor.py
View File

@@ -56,19 +56,10 @@ class GPUExecutor:
distributed_init_method = get_distributed_init_method(
get_ip(), get_open_port())
return Worker(
model_config=self.model_config,
parallel_config=self.parallel_config,
scheduler_config=self.scheduler_config,
device_config=self.device_config,
cache_config=self.cache_config,
load_config=self.load_config,
vllm_config=self.vllm_config,
local_rank=local_rank,
rank=rank,
distributed_init_method=distributed_init_method,
lora_config=self.lora_config,
speculative_config=self.speculative_config,
prompt_adapter_config=self.prompt_adapter_config,
observability_config=self.observability_config,
)
def determine_num_available_blocks(self) -> Tuple[int, int]: