xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-09 22:25:32 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Frontend] Perform offline path replacement to tokenizer (#29706)

Browse Source 
Signed-off-by: Tsukasa OI <floss_llm@irq.a4lg.com>
This commit is contained in:
Tsukasa OI 2025-11-29 11:32:08 +09:00 committed by GitHub
parent b2c50eda50
commit 762a4a6ca9
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 27 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
tests/entrypoints/offline_mode/test_offline_mode.py
View File

@ -23,6 +23,16 @@ MODEL_CONFIGS = [
"max_num_seqs": 64,
"tensor_parallel_size": 1,
},
{
"model": "Qwen/Qwen3-0.6B",
"enforce_eager": True,
"gpu_memory_utilization": 0.50,
"max_model_len": 64,
"max_num_batched_tokens": 64,
"max_num_seqs": 64,
"tensor_parallel_size": 1,
"tokenizer": "Qwen/Qwen3-4B",
},
{
"model": "mistralai/Mistral-7B-Instruct-v0.1",
"enforce_eager": True,

23
vllm/engine/arg_utils.py
View File

@ -581,15 +581,26 @@ class EngineArgs:
from vllm.plugins import load_general_plugins
load_general_plugins()
# when use hf offline,replace model id to local model path
# when use hf offline,replace model and tokenizer id to local model path
if huggingface_hub.constants.HF_HUB_OFFLINE:
model_id = self.model
self.model = get_model_path(self.model, self.revision)
logger.info(
"HF_HUB_OFFLINE is True, replace model_id [%s] to model_path [%s]",
model_id,
self.model,
)
if model_id is not self.model:
logger.info(
"HF_HUB_OFFLINE is True, replace model_id [%s] to model_path [%s]",
model_id,
self.model,
)
if self.tokenizer is not None:
tokenizer_id = self.tokenizer
self.tokenizer = get_model_path(self.tokenizer, self.tokenizer_revision)
if tokenizer_id is not self.tokenizer:
logger.info(
"HF_HUB_OFFLINE is True, replace tokenizer_id [%s] "
"to tokenizer_path [%s]",
tokenizer_id,
self.tokenizer,
)
@staticmethod
def add_cli_args(parser: FlexibleArgumentParser) -> FlexibleArgumentParser: