用回收的随机噪声提升文本生成一致性，效果好且不慢。
Waste Not, Want Not; Recycled Gumbel Noise Improves Consistency in Natural Language Generation
- 用Gumbel重参数化引入潜变量控制生成过程
- 在语义和风格一致性上比标准采样提升最高10%
- 兼容现有方法，计算开销几乎为零
语言模型输出的一致性对其可靠性和实用性至关重要。由于训练目标，语言模型会学习建模所有可能的延续，导致相同或相似输入产生风格与内容差异较大的输出。为此，我们提出一种新型解码算法，在不降低输出质量的前提下，显著提升不同提示下的响应一致性。该方法基于Gumbel重参数化技巧，在下一词采样中引入潜变量，使模型生成更稳定。实验表明，该方法在语义和风格一致性基准上相比标准采样最高提升10%。同时，该方法可无缝集成至现有采样策略，计算开销极低，为提升语言模型输出可靠性提供了实用方案。
原文摘要 · Abstract (English)
Consistency in the output of language models is critical for their reliability and practical utility. Due to their training objective, language models learn to model the full space of possible continuations, leading to outputs that can vary significantly in style and content, even for similar or repeated inputs. To address this, we propose a novel decoding algorithm that enhances response consistency across different prompts with no degradation in response quality. By incorporating a latent variable into the next-token sampling process based on the Gumbel reparametrisation trick, our method outperforms standard sampling by up to 10% across semantic and stylistic consistency benchmarks. Additionally, our approach integrates seamlessly with existing sampling methods with negligible computational overhead, providing a practical solution for improving the reliability of language model outputs.
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。