CriticGen: Generation-Aware Evaluation as Actionable Feedback

CriticGen：作为可操作反馈的生成感知评估

Current evaluation methods for large language models are coarse-grained and decoupled from generation, producing generic explanations that fail to provide actionable feedback for model improvement. 当前大语言模型的评估方法粒度较粗，且与生成过程脱节，产生的解释往往过于通用，无法为模型改进提供可操作的反馈。

We propose CriticGen, a fine-grained, generation-aware evaluation framework that turns evaluation into actionable control for answer improvement. CriticGen first generates sample-specific evaluation dimensions and scoring criteria under high-level categories such as subjective, objective, and self-derived constraints. 我们提出了 CriticGen，这是一个细粒度的、生成感知的评估框架，它将评估转化为改进答案的可操作控制。CriticGen 首先在主观、客观和自导约束等高层类别下，生成针对特定样本的评估维度和评分标准。

These criteria then serve as a dynamic rubric for jointly producing a score, a reason, an executable refinement suggestion, and a refined answer. This rubric-conditioned refinement process enables models to diagnose flaws and perform targeted answer improvement. 这些标准随后作为动态评分准则，用于共同生成评分、理由、可执行的优化建议以及优化后的答案。这种基于准则的优化过程使模型能够诊断缺陷并进行有针对性的答案改进。

Experimental results show that fine-grained evaluation should be both instance-specific and actionable. CriticGen induces higher-quality rubrics, improving relevance/coverage from 3.33/4.03 to 3.97/4.24. 实验结果表明，细粒度评估应当同时具备实例特异性和可操作性。CriticGen 能够引导出更高质量的评分准则，将相关性/覆盖率从 3.33/4.03 提升至 3.97/4.24。

CriticGen also achieves the best score correlations, with 0.9556 Pearson and 0.9560 Spearman, and raises the F1 of criterion-grounded reasons and executable suggestions from 0.6369/0.5994 to 0.7554/0.7900. CriticGen 还实现了最佳的评分相关性，Pearson 相关系数达到 0.9556，Spearman 相关系数达到 0.9560，并将基于准则的理由和可执行建议的 F1 分数分别从 0.6369/0.5994 提升至 0.7554/0.7900。

Crucially, its feedback translates into reliable answer improvement, improving 73.17% of answers with a 93.28% non-degradation rate. 至关重要的是，其反馈能够转化为可靠的答案改进，使 73.17% 的答案得到优化，且非退化率（即不降低质量的比例）高达 93.28%。