RaguTeam at SemEval-2026 Task 8: Meno and Friends in a Judge-Orchestrated LLM Ensemble for Faithful Multi-Turn Response Generation

Hugging Face Daily Papers Papers

Summary

This paper presents the winning system for SemEval-2026 Task 8's generation subtask, using a heterogeneous ensemble of seven LLMs with dual prompting strategies and a GPT-4o-mini judge to select the best response. The system achieved first place with a conditioned harmonic mean of 0.7827, outperforming all baselines and demonstrating the value of model diversity.

We present our winning system for Task~B (generation with reference passages) in SemEval-2026 Task~8: MTRAGEval. Our method is a heterogeneous ensemble of seven LLMs with two prompting variants, where a GPT-4o-mini judge selects the best candidate per instance. We ranked 1st out of 26 teams, achieving a conditioned harmonic mean of 0.7827 and outperforming the strongest baseline (gpt-oss-120b, 0.6390). Ablations show that diversity in model families, scales, and prompting strategies is essential, with the ensemble consistently beating any single model. We also introduce Meno-Lite-0.1, a 7B domain-adapted model with a strong cost--performance trade-off, and analyse MTRAGEval, highlighting annotation limitations and directions for improvement. Our code is publicly available: https://github.com/RaguTeam/ragu_mtrag_semeval
Original Article
View Cached Full Text

Cached at: 05/08/26, 06:52 AM

Paper page - RaguTeam at SemEval-2026 Task 8: Meno and Friends in a Judge-Orchestrated LLM Ensemble for Faithful Multi-Turn Response Generation

Source: https://huggingface.co/papers/2605.04523

Abstract

A heterogeneous ensemble of seven large language models with dual prompting strategies achieved top performance in the SemEval-2026 MTRAGEval task through judge selection and demonstrated the importance of model diversity.

We present our winning system for Task~B (generation with reference passages) in SemEval-2026 Task~8: MTRAGEval. Our method is aheterogeneous ensembleof seven LLMs with twoprompting variants, where aGPT-4o-minijudge selects the best candidate per instance. We ranked 1st out of 26 teams, achieving aconditioned harmonic meanof 0.7827 and outperforming the strongest baseline (gpt-oss-120b, 0.6390). Ablations show that diversity in model families, scales, and prompting strategies is essential, with the ensemble consistently beating any single model. We also introduce Meno-Lite-0.1, a 7Bdomain-adapted modelwith a strongcost--performance trade-off, and analyse MTRAGEval, highlighting annotation limitations and directions for improvement. Our code is publicly available: https://github.com/RaguTeam/ragu_mtrag_semeval

View arXiv pageView PDFGitHub0Add to collection

Models citing this paper1

#### bond005/meno-lite-0.1 Text Generation• 8B• Updatedabout 3 hours ago • 168 • 5

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2605.04523 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2605.04523 in a Space README.md to link it from this page.

Collections including this paper1

Similar Articles

Mixture of Complementary Agents for Robust LLM Ensemble

arXiv cs.LG

Proposes a framework for selecting complementary LLMs as proposers in ensemble systems, reformulating proposer selection as a combinatorial problem and exploring greedy algorithms for efficient performance-cost trade-offs.