Abstract:In-Context Learning has shown great potential for aligning Large Language Models (LLMs) with human values, helping reduce harmful outputs and accommodate diverse preferences without costly post-training, known as In-Context Alignment (ICA). However, LLMs' comprehension of input prompts remains agnostic, limiting ICA's ability to address value tensions--human values are inherently pluralistic, often imposing conflicting demands, e.g., stimulation vs. tradition. Current ICA methods therefore face the Instruction Bottleneck challenge, where LLMs struggle to reconcile multiple intended values within a single prompt, leading to incomplete or biased alignment. To address this, we propose PICACO, a novel pluralistic ICA method. Without fine-tuning, PICACO optimizes a meta-instruction that navigates multiple values to better elicit LLMs' understanding of them and improve their alignment. This is achieved by maximizing the total correlation between specified values and LLM responses, theoretically reinforcing value correlation while reducing distractive noise, resulting in effective value instructions. Extensive experiments on five value sets show that PICACO works well with both black-box and open-source LLMs, outperforms several recent strong baselines, and achieves a better balance across up to 8 distinct values.
| Comments: | ICML 2026 |
| Subjects: | Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computers and Society (cs.CY) |
| Cite as: | arXiv:2507.16679 [cs.CL] |
| (or arXiv:2507.16679v3 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2507.16679 arXiv-issued DOI via DataCite |
From: Han Jiang [view email]
[v1]
Tue, 22 Jul 2025 15:14:56 UTC (1,622 KB)
[v2]
Thu, 29 Jan 2026 22:01:03 UTC (1,634 KB)
[v3]
Tue, 26 May 2026 15:36:45 UTC (1,641 KB)
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。