# LLM-INSTRUCT at UZH Shared Task 2026: Constraint-Aware Retrieval and Selective Debate for Paragraph-Level Argument Mining

arXiv: https://arxiv.org/abs/2607.20430 · HTML: https://mlsanity.com/paper/2607.20430/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 10 of its 14 references that are in the map.
- First posted on arXiv 10 May 2026.

## Builds on (most cited first, 10 of 10)

- [Dense Passage Retrieval for Open-Domain Question Answering](https://mlsanity.com/paper/2004.04906/) (2020)
- [Improving Factuality and Reasoning in Language Models through Multiagent Debate](https://mlsanity.com/paper/2305.14325/) (2023)
- [Text Embeddings by Weakly-Supervised Contrastive Pre-training](https://mlsanity.com/paper/2212.03533/) (2022)
- [Grammar-Constrained Decoding for Structured NLP Tasks without Finetuning](https://mlsanity.com/paper/2305.13971/) (2023)
- [Structured Prediction as Translation between Augmented Natural Languages](https://mlsanity.com/paper/2101.05779/) (2021)
- [Parsing Argumentation Structures in Persuasive Essays](https://mlsanity.com/paper/1604.07370/) (2016)
- [Autoregressive Structured Prediction with Language Models](https://mlsanity.com/paper/2210.14698/) (2022)
- [Can Large Language Models perform Relation-based Argument Mining?](https://mlsanity.com/paper/2402.11243/) (2024)
- [AutoAM: An End-To-End Neural Model for Automatic and Universal Argument Mining](https://mlsanity.com/paper/2309.09300/) (2023)
- [SpiritRAG: A Q&A System for Religion and Spirituality in the United Nations Archive](https://mlsanity.com/paper/2507.04395/) (2025)

## Abstract (arXiv)

We present LLM-INSTRUCT, the winning system for the UZH Shared Task at ArgMining 2026 on paragraph-level argument mining in UN and UNESCO resolutions. The task requires paragraph-type classification, prediction of a subset of 141 official tags, and directed relation prediction under a strict JSON schema setting using only open-weight models up to 8B parameters. We frame the task as constrained structured prediction. The system first narrows the candidate tag space with metadata-aware dense retrieval, then applies constrained decoding with per-dimension caps, escalates only uncertain cases to a three-agent debate branch, and finally validates the output schema. On the official leaderboard, LLM-INSTRUCT ranked 1st overall, with 1st in F1 and 5th in LLM-as-a-Judge. During development, our configuration search further improved Task 1b Micro-F1 from 35.83% to 40.08% while keeping the internal Task 2 score at 4.421. The main lesson is simple: reducing the decision space before generation improves both accuracy and submission robustness. Our code and supporting scripts are publicly available at: https://github.com/LLM-Instruct-at-UZH-Shared-Task-2026/Method