Balancing Label Quantity and Quality for Scalable Elicitation
Keywords: Scalable oversight, Alignment, Safety, Few-shot learning, Eliciting latent knowledge, Weak-to-strong generalization
TL;DR: We compare learning methods and optimally trade off the quantity and quality of labels to elicit knowledge from a pretrained model in a labeling-cost constrained setting.
Abstract: Scalable oversight studies methods of training and evaluating AI systems in domains where human judgement is unreliable or expensive, such as scientific research and software engineering in complex codebases. Recent work in this area by Burns et al. (2023) suggests that Language Models (LMs) pretrained on internet-scale corpora exhibit an inductive bias toward producing correct answers, even when finetuned on error-prone labels produced by a smaller language model. This suggests that massive pretraining combined with finetuning on imperfect human labels may be a solid baseline method for scalable oversight. In the real world, however, label quality is not fixed: practitioners face a quantity-quality tradeoff when generating finetuning data. In this paper, we explore the microeconomics of the quantity-quality tradeoff on binary NLP classification tasks used in Burns et al. (2023). We find that there are three regimes of eliciting classification knowledge from pretrained models using supervised finetuning: quantity-dominant, quality-dominant, and a mixed regime involving the use of low- and high-quality data together to attain higher accuracy at a lower cost than using either alone. We explore sample-efficient elicitation methods that make use of two datasets of differing qualities, and establish a Pareto frontier of scalable elicitation methods that optimally trade off labeling cost and classifier performance.
Primary Area: alignment, fairness, safety, privacy, and societal considerations
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 8270
Loading