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
[Build] Update Xgrammar to 0.1.25 (#25467)
Signed-off-by: chaunceyjiang <chaunceyjiang@gmail.com>
Chauncey committed
Sep 24, 2025 at 05:25 UTC
08275ec0a2ba48fbc1054bdbdda2f1e0dfcb20b3
2 files changed
+7
-3
requirements/common.txt
+1
-1
@@ -24,7 +24,7 @@ outlines_core == 0.2.11
24
# required for outlines backend disk cache
25
diskcache == 5.6.3
26
lark == 1.2.2
27
-xgrammar == 0.1.24; platform_machine == "x86_64" or platform_machine == "aarch64" or platform_machine == "arm64"
27
+xgrammar == 0.1.25; platform_machine == "x86_64" or platform_machine == "aarch64" or platform_machine == "arm64"
28
typing_extensions >= 4.10
29
filelock >= 3.16.1 # need to contain https://github.com/tox-dev/filelock/pull/317
30
partial-json-parser # used for parsing partial JSON outputs
vllm/v1/structured_output/backend_xgrammar.py
+6
-2
@@ -108,7 +108,9 @@ class XgrammarBackend(StructuredOutputBackend):
108
end=s["end"],
109
) for s in s_tag["structures"]
110
]
111
- ctx = self.compiler.compile_structural_tag(tags, s_tag["triggers"])
111
+ structural_tag = xgr.StructuralTag.from_legacy_structural_tag(
112
+ tags, s_tag["triggers"])
113
+ ctx = self.compiler.compile_structural_tag(structural_tag)
114
else:
115
logger.error(
116
"Validation should have already occurred. Please file an issue."
@@ -318,6 +320,8 @@ def validate_xgrammar_grammar(sampling_params: SamplingParams) -> None:
320
end=s["end"],
321
) for s in s_tag["structures"]
322
]
321
- xgr.Grammar.from_structural_tag(tags, s_tag["triggers"])
323
+ structural_tag = xgr.StructuralTag.from_legacy_structural_tag(
324
+ tags, s_tag["triggers"])
325
+ xgr.Grammar.from_structural_tag(structural_tag)
326
except Exception as e:
327
raise ValueError("Invalid structural tag specification.") from e