--- title: "No-Worse Context-Aware Decoding: Preventing Neutral" description: "The paper proposes NWCAD, a decode-time adapter that runs a with-context and a no-context forward… 1 reference, cited by 1 paper here." authors: ["Yufei Tao", "Ameeta Agrawal"] year: 2026 venue: "Findings of the Association for Computational Linguistics: ACL 2026" source: "https://aclanthology.org/2026.findings-acl.473/" doi: "10.18653/v1/2026.findings-acl.473" licence: "CC BY 4.0 (ACL Anthology)" licence_url: "https://creativecommons.org/licenses/by/4.0/" summary_licence: "Elamur Cosmic License" canonical: "https://research.elamur.ai/free/en/papers/tao-2026-no-worse-context-aware-decoding/" --- [Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/tao-2026-no-worse-context-aware-decoding/) [home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2026](https://md.research.elamur.ai/free/en/papers/2026/page/15/) / No-Worse Context-Aware Decoding: Preventing Neutral Regression in Context-Conditioned Generation # No-Worse Context-Aware Decoding: Preventing Neutral Regression in Context-Conditioned Generation **Authors:** Yufei Tao and [Ameeta Agrawal](https://md.research.elamur.ai/free/en/authors/ameeta-agrawal/) · **Year:** 2026 · **Venue:** [Findings of the Association for Computational Linguistics: ACL 2026](https://md.research.elamur.ai/free/en/venues/findings/) ([Findings of the Association for Computational Linguistics 2026](https://md.research.elamur.ai/free/en/venues/findings/2026/)) · **Publisher:** [Association for Computational Linguistics](https://md.research.elamur.ai/free/en/publishers/association-for-computational-linguistics/) · **Licence:** [CC BY 4.0 (ACL Anthology)](https://creativecommons.org/licenses/by/4.0/) **Links:** [Publisher's page](https://aclanthology.org/2026.findings-acl.473/) · [PDF](https://aclanthology.org/2026.findings-acl.473.pdf) · [DOI 10.18653/v1/2026.findings-acl.473](https://doi.org/10.18653/v1/2026.findings-acl.473) · [OpenAlex](https://openalex.org/W7166868610) **Citations:** 0 citations (OpenAlex, 2026-10-14) · [1](https://md.research.elamur.ai/free/en/papers/tao-2026-no-worse-context-aware-decoding/cited-by/) in our library, of the 149,439 papers whose bibliographies we read ## Summary LLMs given external passages, such as retrieved evidence in RAG, sometimes change an answer they already had right, even when the passage adds nothing useful. The authors call this failure "neutral regression". They measure it as the accuracy lost on baseline-correct questions when the supplied context agrees with the answer. Existing contrastive decoders (CAD, AdaCAD, CoCoA) always reweight logits, so they cannot promise to leave such answers untouched. The paper proposes NWCAD, a decode-time adapter that runs a with-context and a no-context forward pass in parallel. At each token, a two-stage gate makes one of three choices: - It copies the no-context logits exactly when both streams agree and the no-context stream is confident. - It uses the with-context logits when that stream is decisive. - Otherwise, it applies a CAD-style fallback. Under greedy decoding, the backoff branch provably reproduces the no-context token. Evaluation uses controlled slices that separate do-no-harm cases from genuinely helpful contexts, plus several QA benchmarks. NWCAD avoids neutral regression while keeping strong accuracy when context helps, adds almost no latency, and needs no extra models. ## Key points - **Gating signals.** Agreement between streams is measured by Jensen–Shannon divergence, approximated over the union of each stream's top-50 tokens. Confidence is the top-1 vs. top-2 probability margin. The thresholds τ, κ_pri and κ_ctx tune how conservative the decoder is. - **Evaluation data.** Controlled slices (Restate-hard, Distractor-hard, Helpful) and benchmarks including NQ-SYNTH, NQ-SWAP, HotpotQA-distractor, ToFuEval and ExpertQA, tested on several backbones including Llama-3.1-8B. - **Main gains.** - Stage 2 adds about 5.2% average accuracy over the Stage 1-only variant. - Wrapping CAD, AdaCAD or CoCoA with NWCAD improves controlled-slice accuracy by roughly 7 to over 40 points, with the largest gains for CAD and CoCoA. - **Fallback rarely needed.** The fallback fires on only about 1–2% of tokens, and removing it barely changes results. Most of the benefit comes from choosing between the no-context and with-context streams. - **Cost and limitations.** Latency stays within about 2% of the base decoder and is sometimes faster. Thresholds were tuned on Llama-3.1-8B only. The method requires logit access, so it does not apply to black-box APIs. Code: github.com/CastGryff/NWCAD. *Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.* ## Institutions - [Portland State University](https://md.research.elamur.ai/free/en/universities/portland-state-university/) — University, Portland, United States · 6 works · 53 citations 📚 Shelves (5): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 16,667 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 16,194 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 16,039 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 15,891 works · [🗄️ Information Systems](https://md.research.elamur.ai/free/en/shelves/computer-science/information-systems/) · 1,437 works ## Cite Cite as: Yufei Tao and Ameeta Agrawal. 2026. No-Worse Context-Aware Decoding: Preventing Neutral Regression in Context-Conditioned Generation. Findings of the Association for Computational Linguistics: ACL 2026. The paper itself, not this page. BibTeX from the library's record of the work. ```bibtex @inproceedings{tao2026no, title = {{No-Worse Context-Aware Decoding: Preventing Neutral Regression in Context-Conditioned Generation}}, author = {Yufei Tao and Ameeta Agrawal}, booktitle = {Findings of the Association for Computational Linguistics: ACL 2026}, year = {2026}, doi = {10.18653/v1/2026.findings-acl.473}, url = {https://doi.org/10.18653/v1/2026.findings-acl.473} } ``` - [← previous No Shortcuts to Culture: Indonesian Multi-hop Question Answering for Complex Cultural Understanding](https://md.research.elamur.ai/free/en/papers/permadi-2026-no-shortcuts-culture-indonesian-multi/) - [next → Not All Tokens Matter: Towards Efficient LLM Reasoning via Token Significance in Reinforcement Learning](https://md.research.elamur.ai/free/en/papers/liu-2026-not-all-tokens-matter-efficient/) - [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/15/) ## References 1 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet. 1. Anant Khandelwal, Manish Gupta, and Puneet Agrawal. 2025. Cocoa: Confidence and context-aware adaptive decoding for resolving knowledge conflicts in large language models. Preprint, arXiv:2508.17670. \#\# Conclusion [find](https://scholar.google.com/scholar?q=Anant+Khandelwal%2C+Manish+Gupta%2C+and+Puneet+Agrawal%2E+2025%2E+Cocoa%3A+Confidence+and+context%2Daware+adaptive+decoding+for+resolving+knowledge+conflicts+in+large+language+models%2E+Preprint%2C+arXiv%3A2508%2E17670%2E+) *Updated 31 October 2026*