pith. sign in

TokenButler: Token Importance is Predictable

2 Pith papers cite this work. Polarity classification is still indexing.

2 Pith papers citing it
abstract

Large Language Models (LLMs) rely on the Key-Value (KV) Cache to store token history, enabling efficient decoding of tokens. As the KV-Cache grows, it becomes a major memory and computation bottleneck. However, there is an opportunity to alleviate this bottleneck, prior research has shown that only a small subset of tokens contribute meaningfully to each decoding step. A key challenge in finding these critical tokens is that they are dynamic, and heavily input query-dependent. Existing methods either risk quality by evicting tokens permanently, or retain the full KV-Cache but rely on retrieving chunks of tokens and many existing KV-Cache sparsity methods rely on inaccurate proxies for token importance. To address these limitations, we introduce TokenButler, a high-granularity, query-aware predictor that learns to identify these critical tokens. TokenButler predicts low-dimensional importance queries at a fixed depth stride, and combines them with a learned projection of the real KV-cache keys to score tokens cheaply, enabling dynamic per-token selection under a fixed budget while preserving the full KV cache. We train TokenButler by distilling the model's masked causal attention distributions, optimizing a lightweight predictor with minimal parameter overhead. We evaluate TokenButler on a novel synthetic small-context co-referential retrieval task, demonstrating near-oracle accuracy where existing methods fail. Furthermore, TokenButler achieves competitive or superior performance on long-context benchmarks (RULER, LongBench), up to $\approx1.6\times$ on-GPU speedup using our proposed *prediction interval with neighbor fetching* that amortizes predictor cost while maintaining accuracy within $\approx$1.1\%, and up to 7.6$\times$ reduction in latency compared to Dense Attention with CPU offloading. Code is available: https://github.com/abdelfattah-lab/TokenButler

citation-role summary

background 1

citation-polarity summary

fields

cs.AI 1 cs.LG 1

years

2026 2

verdicts

UNVERDICTED 2

roles

background 1

polarities

unclear 1

representative citing papers

Compute Where it Counts: Self Optimizing Language Models

cs.LG · 2026-05-11 · unverdicted · novelty 6.0

SOL trains a policy to dynamically control multiple efficiency mechanisms per token via group-relative policy optimization on teacher-forced episodes, yielding better quality at matched average budget than static or random allocation.

citing papers explorer

Showing 2 of 2 citing papers.

  • Compute Where it Counts: Self Optimizing Language Models cs.LG · 2026-05-11 · unverdicted · none · ref 2 · internal anchor

    SOL trains a policy to dynamically control multiple efficiency mechanisms per token via group-relative policy optimization on teacher-forced episodes, yielding better quality at matched average budget than static or random allocation.

  • Computational Challenges in Token Economics: Bridging Economic Theory and AI System Design cs.AI · 2026-05-17 · unverdicted · none · ref 7 · internal anchor

    The paper defines Computational Token Economics and introduces the Token Economics Trilemma as a framework for trade-offs in granularity, real-time performance, and optimality, while outlining a research agenda for three challenge areas.