A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 14, 05:30 UTC
[Bugfix] Fix args settings for guided decoding args (#27375)
Signed-off-by: Lucia Fang <fanglu@fb.com>
Lucia Fang committed
Oct 23, 2025 at 00:34 UTC
fc059c706180e38e8425a4d0f5bd6b7adb1cd7d0
1 file changed
+3
-5
vllm/engine/arg_utils.py
+3
-5
@@ -1600,15 +1600,13 @@ class EngineArgs:
1600
if self.guided_decoding_backend is not None:
1601
so_config.guided_decoding_backend = self.guided_decoding_backend
1602
if self.guided_decoding_disable_fallback is not None:
1603
- so_config.guided_decoding_disable_fallback = (
1604
- self.guided_decoding_disable_fallback
1605
- )
1603
+ so_config.disable_fallback = self.guided_decoding_disable_fallback
1604
if self.guided_decoding_disable_any_whitespace is not None:
1607
- so_config.guided_decoding_disable_any_whitespace = (
1605
+ so_config.disable_any_whitespace = (
1606
self.guided_decoding_disable_any_whitespace
1607
)
1608
if self.guided_decoding_disable_additional_properties is not None:
1611
- so_config.guided_decoding_disable_additional_properties = (
1609
+ so_config.disable_additional_properties = (
1610
self.guided_decoding_disable_additional_properties
1611
)
1612