pith. machine review for the scientific record. sign in

DySCO: Dynamic Attention-Scaling Decoding for Long-Context Language Models

1 Pith paper cite this work. Polarity classification is still indexing.

1 Pith paper citing it
abstract

Understanding and reasoning over long contexts is a crucial capability for language models (LMs). Although recent models support increasingly long context windows, their accuracy often deteriorates as input length grows. In practice, models often struggle to keep attention aligned with the most relevant context throughout decoding. In this work, we propose DYSCO, a novel decoding algorithm for improving long-context reasoning. DYSCO leverages retrieval heads--a subset of attention heads specialized for longcontext retrieval--to identify task-relevant tokens at each decoding step and explicitly up-weight them. By doing so, DYSCO dynamically adjusts attention during generation to better utilize relevant context. The method is training-free and can be applied directly to any off-the-shelf LMs. Across multiple instruction-tuned and reasoning models, DYSCO consistently improves performance on challenging long-context reasoning benchmarks, yielding relative gains of up to 25% on MRCR and LongBenchV2 at 128K context length with modest additional compute. Further analysis highlights the importance of both dynamic attention rescaling and retrievalhead guided selection for the effectiveness of the method, while providing interpretability insights into decoding-time attention behavior. Our code is available at https://github.com/princeton-pli/DySCO.

fields

cs.CL 1

years

2026 1

verdicts

UNVERDICTED 1

representative citing papers

FocuSFT: Bilevel Optimization for Dilution-Aware Long-Context Fine-Tuning

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

FocuSFT uses an inner optimization loop to adapt fast-weight parameters into a parametric memory that sharpens attention on relevant content, then conditions outer-loop supervised fine-tuning on this representation, yielding gains on long-context benchmarks.

citing papers explorer

Showing 1 of 1 citing paper.

  • FocuSFT: Bilevel Optimization for Dilution-Aware Long-Context Fine-Tuning cs.CL · 2026-05-11 · unverdicted · none · ref 39 · internal anchor

    FocuSFT uses an inner optimization loop to adapt fast-weight parameters into a parametric memory that sharpens attention on relevant content, then conditions outer-loop supervised fine-tuning on this representation, yielding gains on long-context benchmarks.