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

[Core] Support Lora lineage and base model metadata management (#6315)

Browse Source
This commit is contained in:
Jiaxin Shan
2024-09-19 23:20:56 -07:00
committed by GitHub
parent 9e5ec35b1f
commit 260d40b5ea
15 changed files with 337 additions and 45 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
vllm/entrypoints/openai/serving_tokenization.py
View File

@@ -16,7 +16,8 @@ from vllm.entrypoints.openai.protocol import (DetokenizeRequest,
TokenizeRequest,
TokenizeResponse)
# yapf: enable
from vllm.entrypoints.openai.serving_engine import (LoRAModulePath,
from vllm.entrypoints.openai.serving_engine import (BaseModelPath,
LoRAModulePath,
OpenAIServing)
from vllm.logger import init_logger
from vllm.transformers_utils.tokenizer import MistralTokenizer
@@ -31,7 +32,7 @@ class OpenAIServingTokenization(OpenAIServing):
self,
engine_client: EngineClient,
model_config: ModelConfig,
served_model_names: List[str],
base_model_paths: List[BaseModelPath],
*,
lora_modules: Optional[List[LoRAModulePath]],
request_logger: Optional[RequestLogger],
@@ -39,7 +40,7 @@ class OpenAIServingTokenization(OpenAIServing):
):
super().__init__(engine_client=engine_client,
model_config=model_config,
served_model_names=served_model_names,
base_model_paths=base_model_paths,
lora_modules=lora_modules,
prompt_adapters=None,
request_logger=request_logger)