Sign up to save your podcasts
Or
Share Active Learners as Efficient PRP Rerankers
Share to email
Share to Facebook
Share to X
Share to email
Share to Facebook
Share to X
Or
Active Learners as Efficient PRP Rerankers
🤗 Upvotes: 85 | cs.LG, cs.AI, cs.CL
Authors:
JeremÃas Figueiredo Paschmann, Juan Kaplan, Francisco Nattero, Santiago Barron, Juan Wisznia, Luciano del Corro
Title:
Active Learners as Efficient PRP Rerankers
Arxiv:
http://arxiv.org/abs/2605.14236v2
Abstract:
Pairwise Ranking Prompting (PRP) elicits pairwise preference judgments from an LLM, which are then aggregated into a ranking, usually via classical sorting algorithms. However, judgments are noisy, order-sensitive, and sometimes intransitive, so sorting assumptions do not match the setting. Because sorting aims to recover a full permutation, truncating it to meet a call budget does not produce a dependable top-K. We thus reframe PRP reranking as active learning from noisy pairwise comparisons and show that active rankers are drop-in replacements that improve NDCG@10 per call in the call-constrained regime. Our noise-robust framework also introduces a randomized-direction oracle that uses a single LLM call per pair. This approach converts systematic position bias into zero-mean noise, enabling unbiased aggregate ranking without the cost of bidirectional calls.
View all episodes
By Jingwen Liang, Gengyu Wang🤗 Upvotes: 85 | cs.LG, cs.AI, cs.CL
Authors:
JeremÃas Figueiredo Paschmann, Juan Kaplan, Francisco Nattero, Santiago Barron, Juan Wisznia, Luciano del Corro
Title:
Active Learners as Efficient PRP Rerankers
Arxiv:
http://arxiv.org/abs/2605.14236v2
Abstract:
Pairwise Ranking Prompting (PRP) elicits pairwise preference judgments from an LLM, which are then aggregated into a ranking, usually via classical sorting algorithms. However, judgments are noisy, order-sensitive, and sometimes intransitive, so sorting assumptions do not match the setting. Because sorting aims to recover a full permutation, truncating it to meet a call budget does not produce a dependable top-K. We thus reframe PRP reranking as active learning from noisy pairwise comparisons and show that active rankers are drop-in replacements that improve NDCG@10 per call in the call-constrained regime. Our noise-robust framework also introduces a randomized-direction oracle that uses a single LLM call per pair. This approach converts systematic position bias into zero-mean noise, enabling unbiased aggregate ranking without the cost of bidirectional calls.