arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2407.21057cs.CLcs.AIcs.LG
收藏

面向长文本生成的多组不确定性量化

Multi-group Uncertainty Quantification for Long-form Text Generation

Carnegie Mellon University(卡内基梅隆大学)

Terrance Liu, Zhiwei Steven Wu

AI总结:

针对长文本生成中现有不确定性量化方法在数据子组上保证失效的问题,引入多校准和多验证共形预测等组条件方法,利用子组信息提升子组内不确定性量化效果且保留全局保证,并构建了相关基准。

摘要:

尽管过往研究已展示了不确定性量化可应用于大语言模型(LLM)输出的方式,但由此得到的不确定性保证在数据子分组中是否仍然成立这一问题仍悬而未决。在本研究中,针对由LLM生成的长文本,我们从两个层面研究不确定性:输出中包含的单个声明层面(通过校准)以及整个输出本身的层面(通过共形预测)。我们以传记生成作为本研究的测试平台,为每段生成文本推导一组(人口统计)属性(例如文本描述的是男性还是女性),以此构成这类数据“子组”。我们发现,尽管两类不确定性量化的经典方法在整个数据集上测量时表现良好,但在检验特定子组时这类保证会失效。在明确这一问题后,我们采用面向组条件的不确定性量化方法——多校准(multicalibration)和多验证共形预测(multivalid conformal prediction),并发现在多种方法中,额外的子组信息始终能改善子组内的校准效果和共形预测效果(同时关键在于保留了整个数据集层面的保证)。由于校准、共形预测及其多组对应方法在长文本生成语境下尚未得到广泛探索,我们认为这些结果为该场景构建了一个基准。

英文摘要:

While past works have shown how uncertainty quantification can be applied to large language model (LLM) outputs, the question of whether resulting uncertainty guarantees still hold within sub-groupings of data remains open. In our work, given some long-form text generated by an LLM, we study uncertainty at both the level of individual claims contained within the output (via calibration) and across the entire output itself (via conformal prediction). Using biography generation as a testbed for this study, we derive a set of (demographic) attributes (e.g., whether some text describes a man or woman) for each generation to form such "subgroups" of data. We find that although canonical methods for both types of uncertainty quantification perform well when measuring across the entire dataset, such guarantees break down when examining particular subgroups. Having established this issue, we invoke group-conditional methods for uncertainty quantification -- multicalibration and multivalid conformal prediction -- and find that across a variety of approaches, additional subgroup information consistently improves calibration and conformal prediction within subgroups (while crucially retaining guarantees across the entire dataset). As the problems of calibration, conformal prediction, and their multi-group counterparts have not been extensively explored in the context of long-form text generation, we consider these results to form a benchmark for this setting.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑