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 2206.13974cs.CL
收藏

面向自然语言生成的生成器-排序器联合学习

Joint Generator-Ranker Learning for Natural Language Generation

Sun Yat-sen University(中山大学);Microsoft Research Asia(微软亚洲研究院);Microsoft Research(微软研究院)

Weizhou Shen, Yeyun Gong, Yelong Shen, Song Wang, Xiaojun Quan, Nan Duan, Weizhu Chen

AI总结:

本文提出JGR联合训练算法,将生成器与排序器集成于单一框架,通过混合目标与对比损失迭代更新,在多个文本生成任务上超越现有方法。

摘要:

先生成后排序是一种广泛使用的文本生成机制,其中生成器产生多个文本候选,排序器在这些文本候选中选择最佳的一个。然而,现有方法通常分别训练生成器和排序器,忽略了可以进一步提升生成质量的相互反馈。为了解决这一局限,我们提出了JGR,一种新颖的联合训练算法,将生成器和排序器集成在单一框架中。JGR使用混合目标优化生成器,该目标结合了数据似然和排序器奖励,并使用对比损失训练排序器,该损失比较生成器的输出。通过迭代更新生成器和排序器,JGR能够有效地协调它们的学习并共同提升它们的质量。我们在各种文本生成任务上评估了JGR,并证明它在三个常见生成场景下的四个公开数据集上超越了现有方法。我们的代码和模型已在 https://github.com/microsoft/ProphetNet/tree/master/JGR 公开提供。

英文摘要:

Generate-then-rank is a widely used mechanism for text generation, where a generator produces multiple text candidates and a ranker chooses the best one among the text candidates. However, existing methods usually train the generator and the ranker individually, neglecting the mutual feedback that could further enhance the generation quality. To tackle this limitation, we propose JGR, a novel joint training algorithm that integrates the generator and the ranker in a single framework. JGR optimizes the generator with a hybrid objective that combines data likelihood and ranker reward, and trains the ranker with a contrastive loss that compares the generator outputs. By iteratively updating the generator and the ranker, JGR can effectively harmonize their learning and enhance their quality jointly. We evaluate JGR on various text generation tasks and demonstrate that it surpasses existing methods on four public datasets across three common generation scenarios. Our code and models are publicly available at https://github.com/microsoft/ProphetNet/tree/master/JGR.

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。

↑