A structured decoding engine that precompiles static grammar templates and uses context-free operators with mask caching to cut time-to-first-token overhead by up to 250x versus XGrammar.
lm-format-enforcer
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
WGRAMMAR: Leverage Prior Knowledge to Accelerate Structured Decoding
A structured decoding engine that precompiles static grammar templates and uses context-free operators with mask caching to cut time-to-first-token overhead by up to 250x versus XGrammar.