AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

When Copying Is Hard: Copy-Constrained Decoding for Exact Span Reproduction

Jinghui Zhang, Lang Gao, Zongfang Liu, Ruihong Zeng, Zirui Song, Rui Yan, Kentaro Inui, Xiuying Chen

NeurIPS 2026 · Large language models › Language, safety and interpretability · approach: Parallel and early-exit decoding · 0 citations (OpenAlex)

Read the paper

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Studies exact verbatim span reproduction from long context in LLMs and proposes a copy-constrained decoding approach to fix within-span drift and boundary errors.

Problem

LLMs often fail to reproduce a target span verbatim from context, drifting within the span or failing to stop at the correct boundary, which matters for tasks like embedding sensitive strings in generated code.

Approach

Constrains decoding to copy from the source context so that generated spans stay exact and terminate at the correct boundary; the abstract does not detail the mechanism.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add CopyGen plugin to frozen LLMsQwen3-4B, Qwen3-8B, Ministral-8B · Exact-copy benchmark, 3 backbones avgEM: 51.9 -> 70.3Improved
add CopyGen on top of SFT backboneLoRA SFT models · Exact-copy benchmark, 3 backbones avgEM: 91.3 -> 95.2Improved
remove margin calibration (ablation)CopyGen · Average of 4 domains, Qwen3-8BEM: 76.97 -> 60.47 when removedImproved
add adaptive hardness weighting in training (ablation)CopyGen · Average of 4 domains, Qwen3-8BEM: 63.67 -> 76.97Improved
add adaptive length proposal instead of fixed length (ablation)CopyGen · Average of 4 domains, Qwen3-8BTokens/s: 354.94 -> 842.10Improved
add adaptive length proposal instead of fixed length (ablation)CopyGen · Average of 4 domains, Qwen3-8BEM: 74.32 -> 76.97Improved
replace learned bias head with fixed margin rules (ablation)CopyGen · Qwen3-8B, 3 real domains avgEM: 72.87 -> 59.45 (margin +1)Hurt
reduce plugin training data to 10% (ablation)CopyGen · Qwen3-8B, 4 domains avgEM: 76.97 -> 70.88Hurt
reduce plugin training data to 50% (ablation)CopyGen · Qwen3-8B, 4 domains avgEM: 76.97 -> 76.69No clear effect
train on other two domains and test on held-out domain (ablation)CopyGen · Qwen3-8B, 3 real domains avgEM: 72.49 -> 65.53 (base 57.97)Mixed
add rule-based constraints (PICARD-style, Longest-Copy)frozen LLM · Exact-copy benchmark, 3 backbones avgEM: 51.85 -> 51.61 (PICARD), 26.32 (Longest-Copy)Hurt
integrate CopyGen in vLLM and SGLangQwen3-8B · 4 domains avgEM: 60.69 -> 77.00 (vLLM)Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy