🛰️ Daily AI Frontier
‹ back to 2026-08-31

Parser States Already Know: Structure-Conditioned KV Persistence for Structured Generation

Research Efficiency & Systems

Ranking

Overall 78
Content 95
Popularity 37

Observed public metrics from 1 member.

Representative image for Parser States Already Know: Structure-Conditioned KV Persistence for Structured Generation

Merged summary

TL;DR - PASK uses parser-state information to preserve schema-critical KV-cache entries during structured LLM generation. On Qwen3-4B, it improves function-calling accuracy under compression while reducing memory use and accelerating serving.

  • Converts parser-derived structural signals into layer-group-specific KV persistence policies.
  • Calibrates policies offline using task-error sensitivity and attention-output distortion, requiring only lightweight lookups online.
  • At a 0.33 target KV budget, surpasses the strongest compressed baseline by an average of 17.39 percentage points across eight BFCL subcategories.
  • Achieves up to 2.2× higher throughput, 3.3× lower time per output token, and 0.53× the peak GPU memory of full KV caching.

Sources (1)

Parser States Already Know: Structure-Conditioned KV Persistence for Structured Generation

arXiv cs.LG Linze Wu, Xinrui Chen 2026-08-28 arXiv:2608.28276
Public signals Semantic Scholar citations 0 · Semantic Scholar influential citations 0
Providers: Hugging Face · N/A OpenAlex · N/A Publisher · N/A Semantic Scholar · Citations 0 · Influential citations 0 X · N/A Fetched 2026-09-18 14:17:57.853624 UTC

TL;DR - PASK uses parser-state information to preserve schema-critical KV-cache entries during structured LLM generation. On Qwen3-4B, it improves function-calling accuracy under compression while reducing memory use and accelerating serving.

  • Converts parser-derived structural signals into layer-group-specific KV persistence policies.
  • Calibrates policies offline using task-error sensitivity and attention-output distortion, requiring only lightweight lookups online.
  • At a 0.33 target KV budget, surpasses the strongest compressed baseline by an average of 17.39 percentage points across eight BFCL subcategories.
  • Achieves up to 2.2× higher throughput, 3.3× lower time per output token, and 0.53× the peak GPU memory of full KV caching.
item →