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

Fix early CUDA init via get_architecture_class_name import (#3770)

Browse Source 
Signed-off-by: Lei Wen <wenlei03@qiyi.com>
Co-authored-by: Lei Wen <wenlei03@qiyi.com>
This commit is contained in:
leiwen83
2024-04-03 02:56:26 +08:00
committed by GitHub
parent 205b94942e
commit ad6eca408b
1 changed files with 2 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
vllm/engine/llm_engine.py
View File

@@ -13,7 +13,6 @@ from vllm.engine.ray_utils import initialize_ray_cluster
from vllm.executor.executor_base import ExecutorBase from vllm.executor.executor_base import ExecutorBase
from vllm.logger import init_logger from vllm.logger import init_logger
from vllm.lora.request import LoRARequest from vllm.lora.request import LoRARequest
from vllm.model_executor.model_loader import get_architecture_class_name
from vllm.outputs import RequestOutput from vllm.outputs import RequestOutput
from vllm.sampling_params import SamplingParams from vllm.sampling_params import SamplingParams
from vllm.sequence import (MultiModalData, SamplerOutput, Sequence, from vllm.sequence import (MultiModalData, SamplerOutput, Sequence,
@@ -115,6 +114,8 @@ class LLMEngine:
# If usage stat is enabled, collect relevant info. # If usage stat is enabled, collect relevant info.
if is_usage_stats_enabled(): if is_usage_stats_enabled():
from vllm.model_executor.model_loader import (
get_architecture_class_name)
usage_message.report_usage( usage_message.report_usage(
get_architecture_class_name(model_config), get_architecture_class_name(model_config),
usage_context, usage_context,