{"slug": "enhancing-small-language-models-for-power-outage-report-generation-via-minimum", "title": "Enhancing Small Language Models for Power Outage Report Generation via Minimum Risk Training", "summary": "Applying Minimum Risk Training to power outage report generation raised Qwen2.5-7B-Instruct's overall accuracy from 16.20% to 68.95%, according to an arXiv paper (2609.27197v1) on the Outage Data Initiative Nationwide (ODIN). The approach optimizes sequence-level evaluation metrics rather than token-level maximum-likelihood objectives, transforming heterogeneous reports into standardized XML compliant with CIM IEC 61968-3. The authors present the result as evidence that sequence-level optimization is effective for domain-specific structured generation.", "body_md": "arXiv:2609.27197v1 Announce Type: new \nAbstract: Minimum Risk Training (MRT) enables neural machine translation models to directly optimize sequence-level evaluation metrics instead of relying only on token- level maximum-likelihood objectives Shen et al. [2016]. Although introduced a decade ago, recent work shows renewed potential for risk-based optimization in modern language models Yang et al. [2024], Jinnai et al. [2025]. We apply MRT to power outage report generation for the Outage Data Initiative Nationwide (ODIN), transforming heterogeneous reports into standardized XML compliant with CIM IEC 61968-3. Our MRT approach improves Qwen2.5-7B-Instruct overall accuracy from 16.20% to 68.95%, demonstrating the effectiveness of sequence- level optimization for domain-specific structured generation", "url": "https://wpnews.pro/news/enhancing-small-language-models-for-power-outage-report-generation-via-minimum", "canonical_source": "https://arxiv.org/abs/2609.27197", "published_at": "2026-09-24 04:00:00+00:00", "updated_at": "2026-09-24 04:30:21.906822+00:00", "lang": "en", "topics": ["natural-language-processing", "large-language-models", "machine-learning", "ai-research"], "entities": ["Qwen2.5-7B-Instruct", "Outage Data Initiative Nationwide", "CIM IEC 61968-3", "arXiv"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/enhancing-small-language-models-for-power-outage-report-generation-via-minimum", "markdown": "https://wpnews.pro/news/enhancing-small-language-models-for-power-outage-report-generation-via-minimum.md", "text": "https://wpnews.pro/news/enhancing-small-language-models-for-power-outage-report-generation-via-minimum.txt", "jsonld": "https://wpnews.pro/news/enhancing-small-language-models-for-power-outage-report-generation-via-minimum.jsonld"}}