无需真实标签，通过弱校准模型提升强模型的可靠性。
Calibration without Ground Truth
- 用弱校准模型作为参考，优化强模型的输出概率分布。
- 在多种大模型上验证，显著降低校准误差和合理损失。
- 适用于无标签场景，适合追求模型可信度的研究者。
Villalobos 等人 [2024] 预测，公开可用的人类文本将在未来十年内耗尽。因此，缺乏真实标签的情况下仍能改进模型变得愈发重要。本文提出一种无需标签的后处理框架，利用一个较弱但更校准的参考模型来改进一个强大但校准不佳的主模型。该框架在任意合理损失函数下保证严格性能提升。其核心思想是：当主模型与参考模型互不校准时，严格改进才可能实现。我们形式化了这一条件，并将其与经济学中的套利与无交易定理相联系，设计了一种高效的 Bregman 投影算法，可在无标签条件下保证最坏情况下的损失下降。在不同规模的代表性 LLM 上的实验表明，该方法显著降低了合理损失与校准误差，性能可媲美有监督基线。
原文摘要 · Abstract (English)
Villalobos et al. [2024] predict that publicly available human text will be exhausted within the next decade. Thus, improving models without access to ground-truth labels becomes increasingly important. We propose a label-free post-processing framework that improves a strong but miscalibrated model using a weaker yet better-calibrated reference. Our framework guarantees a strict performance improvement under any proper loss. Our approach is based on a characterization of when strict improvement is possible: when the strong and reference models are not mutually calibrated. We formalize this condition, connect it to arbitrage and no-trade results from economics, and develop an efficient Bregman projection algorithm that guarantees worst-case loss reduction without labels. Experiments on representative LLMs across varying scales demonstrate that our label-free method significantly reduces proper losses and calibration errors, achieving performance competitive with supervised baselines.
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。