[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"item-2394":3},{"id":4,"title":5,"url":6,"summary":7,"summary_zh":8,"content":8,"source_name":9,"source_url":8,"published_at":10,"category":11,"cover_url":8,"hotness":12,"is_selected":13,"score":14,"score_detail":15,"sources":22,"tags":24,"view_count":30,"doi":8,"paper":31,"created_at":40},2394,"[预印本]Vision-language models know more about agriculture than they show and rubric-grounded verifications close the gap","https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.09417","arXiv 2609.09417（2026-09-08）。基于116个数据集、834个类别、8324张图像基准测试，发现VLM视觉编码器已能编码与DINOv3基线相当可分离性的农业特征；通过Probabilistic Pivot Tournament（PPT）验证器结构化测试时推理，评判F1近两倍于下限，其中Gemma 4 E4B-it's病害F1达0.71，高于其自身上限0.60。但置信度分数与正确性负相关。",null,"arXiv | 2026-09-08","2026-09-08T00:00:00Z","论文",10,false,77,{"impact":16,"substance":17,"depth":16,"authority":18,"freshness":19,"relevant":20,"comment":21},18,22,13,6,1,"预印本以116数据集、8324张图像的规模系统评测农业视觉语言模型，并提出PPT验证器显著提升病害识别F1，方法新颖、数据扎实，对农业AI落地有参考价值，但尚未经同行评审。",[23],{"name":9,"url":6},[25,26,27,28,29],"智慧农业","农业人工智能","视觉语言模型","病害识别","农业基准数据集",0,{"doi":8,"openalex_id":8,"authors":32,"venue":8,"cited_by_count":30,"oa_url":8,"card":33,"direction":37,"ingested_from":39},[],{"tldr":34,"method":35,"finding":36,"direction":37,"opportunity":38},"构建农业图像基准，发现视觉语言模型已编码农业特征，但输出未充分体现。","116个数据集、8324张图像基准测试，用PPT验证器做结构化测试时推理。","PPT评判F1近两倍于下限，Gemma 4病害F1达0.71，但置信度与正确性负相关。","农业人工智能与决策模型","可研究农业VLM置信度校准与不确定性量化，提升病害诊断可靠性。","agent","2026-09-14T00:06:32.815855Z"]