Guardado en:
Detalles Bibliográficos
Autores principales: Li, Linzhang, Dong, Yixin, Wang, Guanjie, Xu, Ziyi, Jiang, Alexander, Chen, Tianqi
Formato: Preprint
Publicado: 2026
Materias:
Acceso en línea:https://arxiv.org/abs/2601.04426
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866910256085336064
author Li, Linzhang
Dong, Yixin
Wang, Guanjie
Xu, Ziyi
Jiang, Alexander
Chen, Tianqi
author_facet Li, Linzhang
Dong, Yixin
Wang, Guanjie
Xu, Ziyi
Jiang, Alexander
Chen, Tianqi
contents Modern LLM agents increasingly rely on dynamic structured generation, such as tool calling and response protocols. Unlike traditional structured generation with static structures, these workloads vary both across requests and within a request, posing new challenges to existing engines. We present XGrammar-2, a structured generation engine for dynamic agentic workloads. Our design is based on two key ideas: first-class support for tag-triggered structure switching, and fine-grained reuse across requests with different output structures. Concretely, XGrammar-2 introduces TagDispatch for dynamic structural dispatching and Cross-Grammar Cache for substructure-level cache reuse across grammars. It further improves efficiency with an Earley-based adaptive token mask cache, just-in-time compilation, and repetition state compression. Experiments show that XGrammar-2 achieves over 6x faster compilation than prior structured generation engines, and incurs near-zero end-to-end overhead in modern LLM serving systems.
format Preprint
id arxiv_https___arxiv_org_abs_2601_04426
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle XGrammar-2: Efficient Dynamic Structured Generation Engine for Agentic LLMs
Li, Linzhang
Dong, Yixin
Wang, Guanjie
Xu, Ziyi
Jiang, Alexander
Chen, Tianqi
Artificial Intelligence
I.2.11
Modern LLM agents increasingly rely on dynamic structured generation, such as tool calling and response protocols. Unlike traditional structured generation with static structures, these workloads vary both across requests and within a request, posing new challenges to existing engines. We present XGrammar-2, a structured generation engine for dynamic agentic workloads. Our design is based on two key ideas: first-class support for tag-triggered structure switching, and fine-grained reuse across requests with different output structures. Concretely, XGrammar-2 introduces TagDispatch for dynamic structural dispatching and Cross-Grammar Cache for substructure-level cache reuse across grammars. It further improves efficiency with an Earley-based adaptive token mask cache, just-in-time compilation, and repetition state compression. Experiments show that XGrammar-2 achieves over 6x faster compilation than prior structured generation engines, and incurs near-zero end-to-end overhead in modern LLM serving systems.
title XGrammar-2: Efficient Dynamic Structured Generation Engine for Agentic LLMs
topic Artificial Intelligence
I.2.11
url https://arxiv.org/abs/2601.04426