xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-05-07 13:25:46 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Bugfix] Guided decoding falls back to outlines when fails to import xgrammar (#12976)

Browse Source 
Signed-off-by: Yuan Tang <terrytangyuan@gmail.com>
This commit is contained in:
Yuan Tang 2025-02-11 13:17:44 -05:00 committed by GitHub
parent deb6c1c6b4
commit 14ecab5be2
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 11 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
vllm/model_executor/guided_decoding/__init__.py
View File

@ -40,6 +40,8 @@ def maybe_backend_fallback(
guided_params.backend = "outlines" guided_params.backend = "outlines"
if guided_params.backend == "xgrammar": if guided_params.backend == "xgrammar":
from vllm.model_executor.guided_decoding.xgrammar_decoding import (
xgr_installed)
# xgrammar only has x86 wheels for linux, fallback to outlines # xgrammar only has x86 wheels for linux, fallback to outlines
from vllm.platforms import current_platform from vllm.platforms import current_platform
if current_platform.get_cpu_architecture() is not CpuArchEnum.X86: if current_platform.get_cpu_architecture() is not CpuArchEnum.X86:
@ -77,6 +79,13 @@ def maybe_backend_fallback(
"Falling back to use outlines instead.") "Falling back to use outlines instead.")
guided_params.backend = "outlines" guided_params.backend = "outlines"
# If the xgrammar module cannot be imported successfully,
# we should still allow users to use guided decoding with a fallback.
elif not xgr_installed:
logger.warning("xgrammar module cannot be imported successfully. "
"Falling back to use outlines instead.")
guided_params.backend = "outlines"
if (guided_params.backend == "outlines" if (guided_params.backend == "outlines"
and guided_params.json_object is not None): and guided_params.json_object is not None):
# outlines doesn't support json_object, fallback to xgrammar # outlines doesn't support json_object, fallback to xgrammar

2
vllm/model_executor/guided_decoding/xgrammar_decoding.py
View File

@ -14,7 +14,9 @@ from transformers import PreTrainedTokenizerFast
try: try:
import xgrammar as xgr import xgrammar as xgr
from xgrammar.base import _core as xgr_core from xgrammar.base import _core as xgr_core
xgr_installed = True
except ImportError: except ImportError:
xgr_installed = False
pass pass
from vllm.model_executor.guided_decoding.utils import (convert_lark_to_gbnf, from vllm.model_executor.guided_decoding.utils import (convert_lark_to_gbnf,