arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2305.19230cs.CLcs.AI
收藏

基于隐藏表示变换的受控文本生成

Controlled Text Generation with Hidden Representation Transformations

University of California, Los Angeles(加州大学洛杉矶分校);Amazon Alexa AI(亚马逊Alexa人工智能);University of Southern California(南加州大学)

Vaibhav Kumar, Hana Koorehdavoudi, Masud Moshtaghi, Amita Misra, Ankit Chadha, Emilio Ferrara

AI总结:

本文提出 CHRT 框架,通过学习隐藏表示变换并结合对比学习实现多属性受控文本生成,在去毒化、情感引导和文本简化任务中优于七个基线,且推理延迟仅增加 0.01 秒。

摘要:

我们提出 CHRT(Control Hidden Representation Transformation,控制隐藏表示变换)——一种受控语言生成框架,用于引导大型语言模型生成与特定属性(如有毒性)相关的文本。CHRT 通过学习到的变换修改基础模型的隐藏表示来获得属性控制。我们采用对比学习框架来学习这些变换,并且这些变换可以组合起来以实现多属性控制。通过在三个属性上与七个基线进行比较,实验证明了 CHRT 的有效性。CHRT 在去毒化、积极情感引导和文本简化任务中优于所有基线,同时最大限度地减少语言质量的损失。此外,我们的方法具有最低的推理延迟,仅比基础模型多 0.01 秒,使其最适合高性能生产环境。我们开源了代码并发布了两个新数据集,以进一步推动受控语言生成研究。

英文摘要:

We propose CHRT (Control Hidden Representation Transformation) - a controlled language generation framework that steers large language models to generate text pertaining to certain attributes (such as toxicity). CHRT gains attribute control by modifying the hidden representation of the base model through learned transformations. We employ a contrastive-learning framework to learn these transformations that can be combined to gain multi-attribute control. The effectiveness of CHRT is experimentally shown by comparing it with seven baselines over three attributes. CHRT outperforms all the baselines in the task of detoxification, positive sentiment steering, and text simplification while minimizing the loss in linguistic qualities. Further, our approach has the lowest inference latency of only 0.01 seconds more than the base model, making it the most suitable for high-performance production environments. We open-source our code and release two novel datasets to further propel controlled language generation research.

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。

↑