提出新方法提升生成模型质量估计的准确性。
Are Generative Models Underconfident? Better Quality Estimation with Boosted Model Probability
- 通过增强多正确选项时的模型置信度来改进质量估计。
- 在不同场景下使相关性提升0.194，优于原始概率。
- 无需增加复杂度，效果接近或超过昂贵的监督方法。
质量估计（QE）是在无真实标签情况下评估生成模型输出质量的方法。最简单的方式是直接使用模型输出概率，但我们发现文本生成模型的输出概率可能偏低。在每个生成步骤中可能存在多个正确选项，导致概率分布更分散，因此低概率不等于低质量。基于此，我们提出一种名为BoostedProb的质量估计方法，在存在多个合理输出选项时增强模型置信度。该方法无需增加计算复杂度，在多种设置下显著优于原始模型概率，平均提升0.194的皮尔逊相关系数；在某些场景下表现接近或超越更复杂的监督或集成方法。
原文摘要 · Abstract (English)
Quality Estimation (QE) is estimating quality of the model output during inference when the ground truth is not available. Deriving output quality from the models' output probability is the most trivial and low-effort way. However, we show that the output probability of text-generation models can appear underconfident. At each output step, there can be multiple correct options, making the probability distribution spread out more. Thus, lower probability does not necessarily mean lower output quality. Due to this observation, we propose a QE approach called BoostedProb, which boosts the model's confidence in cases where there are multiple viable output options. With no increase in complexity, BoostedProb is notably better than raw model probability in different settings, achieving on average +0.194 improvement in Pearson correlation to ground-truth quality. It also comes close to or outperforms more costly approaches like supervised or ensemble-based QE in certain settings.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。